Currently the sve-ptrace test for setting and reading FPSIMD data assumes that the child will start off in FPSIMD only mode and that it can use this to read some FPSIMD mode SVE ptrace data, skipping the test if it can't. This isn't an assumption guaranteed by the ABI and also limits how we can use this testcase within the program. Instead skip the initial read and just generate a FPSIMD format buffer for the write part of the test, making the coverage more robust in the face of future kernel and test program changes.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Shuah Khan skhan@linuxfoundation.org --- tools/testing/selftests/arm64/fp/sve-ptrace.c | 39 ++++++++----------- 1 file changed, 16 insertions(+), 23 deletions(-)

diff --git a/tools/testing/selftests/arm64/fp/sve-ptrace.c b/tools/testing/selftests/arm64/fp/sve-ptrace.c index 7682798adbba..8f6146d89ca4 100644 --- a/tools/testing/selftests/arm64/fp/sve-ptrace.c +++ b/tools/testing/selftests/arm64/fp/sve-ptrace.c @@ -46,7 +46,7 @@ static const struct vec_type vec_types[] = {

#define VL_TESTS (((SVE_VQ_MAX - SVE_VQ_MIN) + 1) * 3) #define FLAG_TESTS 2 -#define FPSIMD_TESTS 3 +#define FPSIMD_TESTS 2

#define EXPECTED_TESTS ((VL_TESTS + FLAG_TESTS + FPSIMD_TESTS) * ARRAY_SIZE(vec_types))

@@ -240,28 +240,24 @@ static void check_u32(unsigned int vl, const char *reg, /* Access the FPSIMD registers via the SVE regset */ static void ptrace_sve_fpsimd(pid_t child, const struct vec_type *type) { - void *svebuf = NULL; - size_t svebufsz = 0; + void *svebuf; struct user_sve_header *sve; struct user_fpsimd_state *fpsimd, new_fpsimd; unsigned int i, j; unsigned char *p; + int ret;

- /* New process should start with FPSIMD registers only */ - sve = get_sve(child, type, &svebuf, &svebufsz); - if (!sve) { - ksft_test_result_fail("get_sve(%s): %s\n", - type->name, strerror(errno)); - + svebuf = malloc(SVE_PT_SIZE(0, SVE_PT_REGS_FPSIMD)); + if (!svebuf) { + ksft_test_result_fail("Failed to allocate FPSIMD buffer\n"); return; - } else { - ksft_test_result_pass("get_sve(%s FPSIMD)\n", type->name); }

- ksft_test_result((sve->flags & SVE_PT_REGS_MASK) == SVE_PT_REGS_FPSIMD, - "Got FPSIMD registers via %s\n", type->name); - if ((sve->flags & SVE_PT_REGS_MASK) != SVE_PT_REGS_FPSIMD) - goto out; + memset(svebuf, 0, SVE_PT_SIZE(0, SVE_PT_REGS_FPSIMD)); + sve = svebuf; + sve->flags = SVE_PT_REGS_FPSIMD; + sve->size = SVE_PT_SIZE(0, SVE_PT_REGS_FPSIMD); + sve->vl = 16; /* We don't care what the VL is */

/* Try to set a known FPSIMD state via PT_REGS_SVE */ fpsimd = (struct user_fpsimd_state *)((char *)sve + @@ -273,12 +269,11 @@ static void ptrace_sve_fpsimd(pid_t child, const struct vec_type *type) p[j] = j; }

- if (set_sve(child, type, sve)) { - ksft_test_result_fail("set_sve(%s FPSIMD): %s\n", - type->name, strerror(errno)); - + ret = set_sve(child, type, sve); + ksft_test_result(ret == 0, "%s FPSIMD set via SVE: %d\n", + type->name, ret); + if (ret) goto out; - }

/* Verify via the FPSIMD regset */ if (get_fpsimd(child, &new_fpsimd)) { @@ -548,11 +543,9 @@ static int do_parent(pid_t child) if (getauxval(vec_types[i].hwcap_type) & vec_types[i].hwcap) { ptrace_sve_fpsimd(child, &vec_types[i]); } else { - ksft_test_result_skip("%s FPSIMD get via SVE\n", - vec_types[i].name); ksft_test_result_skip("%s FPSIMD set via SVE\n", vec_types[i].name); - ksft_test_result_skip("%s set read via FPSIMD\n", + ksft_test_result_skip("%s FPSIMD read\n", vec_types[i].name); }

Provide ABI documentation for SME similar to that for SVE. Due to the very large overlap around streaming SVE mode in both implementation and interfaces documentation for streaming mode SVE is added to the SVE document rather than the SME one.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- Documentation/arm64/index.rst | 1 + Documentation/arm64/sme.rst | 428 ++++++++++++++++++++++++++++++++++ Documentation/arm64/sve.rst | 70 +++++- 3 files changed, 489 insertions(+), 10 deletions(-) create mode 100644 Documentation/arm64/sme.rst

diff --git a/Documentation/arm64/index.rst b/Documentation/arm64/index.rst index 4f840bac083e..ae21f8118830 100644 --- a/Documentation/arm64/index.rst +++ b/Documentation/arm64/index.rst @@ -21,6 +21,7 @@ ARM64 Architecture perf pointer-authentication silicon-errata + sme sve tagged-address-abi tagged-pointers diff --git a/Documentation/arm64/sme.rst b/Documentation/arm64/sme.rst new file mode 100644 index 000000000000..8ba677b87e90 --- /dev/null +++ b/Documentation/arm64/sme.rst @@ -0,0 +1,428 @@ +=================================================== +Scalable Matrix Extension support for AArch64 Linux +=================================================== + +This document outlines briefly the interface provided to userspace by Linux in +order to support use of the ARM Scalable Matrix Extension (SME). + +This is an outline of the most important features and issues only and not +intended to be exhaustive. It should be read in conjunction with the SVE +documentation in sve.rst which provides details on the Streaming SVE mode +included in SME. + +This document does not aim to describe the SME architecture or programmer's +model. To aid understanding, a minimal description of relevant programmer's +model features for SME is included in Appendix A. + + +1. General +----------- + +* PSTATE.SM, PSTATE.ZA, the streaming mode vector length, the ZA + register state and TPIDR2_EL0 are tracked per thread. + +* The presence of SME is reported to userspace via HWCAP2_SME in the aux vector + AT_HWCAP2 entry. Presence of this flag implies the presence of the SME + instructions and registers, and the Linux-specific system interfaces + described in this document. SME is reported in /proc/cpuinfo as "sme". + +* Support for the execution of SME instructions in userspace can also be + detected by reading the CPU ID register ID_AA64PFR1_EL1 using an MRS + instruction, and checking that the value of the SME field is nonzero. [3] + + It does not guarantee the presence of the system interfaces described in the + following sections: software that needs to verify that those interfaces are + present must check for HWCAP2_SME instead. + +* There are a number of optional SME features, presence of these is reported + through AT_HWCAP2 through: + + HWCAP2_SME_I16I64 + HWCAP2_SME_F64F64 + HWCAP2_SME_I8I32 + HWCAP2_SME_F16F32 + HWCAP2_SME_B16F32 + HWCAP2_SME_F32F32 + HWCAP2_SME_FA64 + + This list may be extended over time as the SME architecture evolves. + + These extensions are also reported via the CPU ID register ID_AA64SMFR0_EL1, + which userspace can read using an MRS instruction. See elf_hwcaps.txt and + cpu-feature-registers.txt for details. + +* Debuggers should restrict themselves to interacting with the target via the + NT_ARM_SVE, NT_ARM_SSVE and NT_ARM_ZA regsets. The recommended way + of detecting support for these regsets is to connect to a target process + first and then attempt a + + ptrace(PTRACE_GETREGSET, pid, NT_ARM_<regset>, &iov). + +* Whenever ZA register values are exchanged in memory between userspace and + the kernel, the register value is encoded in memory as a series of horizontal + vectors from 0 to VL/8-1 stored in the same endianness invariant format as is + used for SVE vectors. + +* On thread creation TPIDR2_EL0 is preserved unless CLONE_SETTLS is specified, + in which case it is set to 0. + +2. Vector lengths +------------------ + +SME defines a second vector length similar to the SVE vector length which is +controls the size of the streaming mode SVE vectors and the ZA matrix array. +The ZA matrix is square with each side having as many bytes as a streaming +mode SVE vector. + + +3. Sharing of streaming and non-streaming mode SVE state +--------------------------------------------------------- + +It is implementation defined which if any parts of the SVE state are shared +between streaming and non-streaming modes. When switching between modes +via software interfaces such as ptrace if no register content is provided as +part of switching no state will be assumed to be shared and everything will +be zeroed. + + +4. System call behaviour +------------------------- + +* On syscall PSTATE.ZA is preserved, if PSTATE.ZA==1 then the contents of the + ZA matrix are preserved. + +* On syscall PSTATE.SM will be cleared and the SVE registers will be handled + as per the standard SVE ABI. + +* Neither the SVE registers nor ZA are used to pass arguments to or receive + results from any syscall. + +* On process creation (eg, clone()) the newly created process will have + PSTATE.SM cleared. + +* All other SME state of a thread, including the currently configured vector + length, the state of the PR_SME_VL_INHERIT flag, and the deferred vector + length (if any), is preserved across all syscalls, subject to the specific + exceptions for execve() described in section 6. + + +5. Signal handling +------------------- + +* Signal handlers are invoked with streaming mode and ZA disabled. + +* A new signal frame record za_context encodes the ZA register contents on + signal delivery. [1] + +* The signal frame record for ZA always contains basic metadata, in particular + the thread's vector length (in za_context.vl). + +* The ZA matrix may or may not be included in the record, depending on + the value of PSTATE.ZA. The registers are present if and only if: + za_context.head.size >= ZA_SIG_CONTEXT_SIZE(sve_vq_from_vl(za_context.vl)) + in which case PSTATE.ZA == 1. + +* If matrix data is present, the remainder of the record has a vl-dependent + size and layout. Macros ZA_SIG_* are defined [1] to facilitate access to + them. + +* The matrix is stored as a series of horizontal vectors in the same format as + is used for SVE vectors. + +* If the ZA context is too big to fit in sigcontext.__reserved[], then extra + space is allocated on the stack, an extra_context record is written in + __reserved[] referencing this space. za_context is then written in the + extra space. Refer to [1] for further details about this mechanism. + + +5. Signal return +----------------- + +When returning from a signal handler: + +* If there is no za_context record in the signal frame, or if the record is + present but contains no register data as described in the previous section, + then ZA is disabled. + +* If za_context is present in the signal frame and contains matrix data then + PSTATE.ZA is set to 1 and ZA is populated with the specified data. + +* The vector length cannot be changed via signal return. If za_context.vl in + the signal frame does not match the current vector length, the signal return + attempt is treated as illegal, resulting in a forced SIGSEGV. + + +6. prctl extensions +-------------------- + +Some new prctl() calls are added to allow programs to manage the SME vector +length: + +prctl(PR_SME_SET_VL, unsigned long arg) + + Sets the vector length of the calling thread and related flags, where + arg == vl | flags. Other threads of the calling process are unaffected. + + vl is the desired vector length, where sve_vl_valid(vl) must be true. + + flags: + + PR_SME_VL_INHERIT + + Inherit the current vector length across execve(). Otherwise, the + vector length is reset to the system default at execve(). (See + Section 9.) + + PR_SME_SET_VL_ONEXEC + + Defer the requested vector length change until the next execve() + performed by this thread. + + The effect is equivalent to implicit execution of the following + call immediately after the next execve() (if any) by the thread: + + prctl(PR_SME_SET_VL, arg & ~PR_SME_SET_VL_ONEXEC) + + This allows launching of a new program with a different vector + length, while avoiding runtime side effects in the caller. + + Without PR_SME_SET_VL_ONEXEC, the requested change takes effect + immediately. + + + Return value: a nonnegative on success, or a negative value on error: + EINVAL: SME not supported, invalid vector length requested, or + invalid flags. + + + On success: + + * Either the calling thread's vector length or the deferred vector length + to be applied at the next execve() by the thread (dependent on whether + PR_SME_SET_VL_ONEXEC is present in arg), is set to the largest value + supported by the system that is less than or equal to vl. If vl == + SVE_VL_MAX, the value set will be the largest value supported by the + system. + + * Any previously outstanding deferred vector length change in the calling + thread is cancelled. + + * The returned value describes the resulting configuration, encoded as for + PR_SME_GET_VL. The vector length reported in this value is the new + current vector length for this thread if PR_SME_SET_VL_ONEXEC was not + present in arg; otherwise, the reported vector length is the deferred + vector length that will be applied at the next execve() by the calling + thread. + + * Changing the vector length causes all of ZA, P0..P15, FFR and all bits of + Z0..Z31 except for Z0 bits [127:0] .. Z31 bits [127:0] to become + unspecified, including both streaming and non-streaming SVE state. + Calling PR_SME_SET_VL with vl equal to the thread's current vector + length, or calling PR_SME_SET_VL with the PR_SVE_SET_VL_ONEXEC flag, + does not constitute a change to the vector length for this purpose. + + * Changing the vector length causes PSTATE.ZA and PSTATE.SM to be cleared. + Calling PR_SME_SET_VL with vl equal to the thread's current vector + length, or calling PR_SME_SET_VL with the PR_SVE_SET_VL_ONEXEC flag, + does not constitute a change to the vector length for this purpose. + + +prctl(PR_SME_GET_VL) + + Gets the vector length of the calling thread. + + The following flag may be OR-ed into the result: + + PR_SME_VL_INHERIT + + Vector length will be inherited across execve(). + + There is no way to determine whether there is an outstanding deferred + vector length change (which would only normally be the case between a + fork() or vfork() and the corresponding execve() in typical use). + + To extract the vector length from the result, bitwise and it with + PR_SME_VL_LEN_MASK. + + Return value: a nonnegative value on success, or a negative value on error: + EINVAL: SME not supported. + + +7. ptrace extensions +--------------------- + +* A new regset NT_ARM_SSVE is defined for access to streaming mode SVE + state via PTRACE_GETREGSET and PTRACE_SETREGSET, this is documented in + sve.rst. + +* A new regset NT_ARM_ZA is defined for ZA state for access to ZA state via + PTRACE_GETREGSET and PTRACE_SETREGSET. + + Refer to [2] for definitions. + +The regset data starts with struct user_za_header, containing: + + size + + Size of the complete regset, in bytes. + This depends on vl and possibly on other things in the future. + + If a call to PTRACE_GETREGSET requests less data than the value of + size, the caller can allocate a larger buffer and retry in order to + read the complete regset. + + max_size + + Maximum size in bytes that the regset can grow to for the target + thread. The regset won't grow bigger than this even if the target + thread changes its vector length etc. + + vl + + Target thread's current streaming vector length, in bytes. + + max_vl + + Maximum possible streaming vector length for the target thread. + + flags + + Zero or more of the following flags, which have the same + meaning and behaviour as the corresponding PR_SET_VL_* flags: + + SME_PT_VL_INHERIT + + SME_PT_VL_ONEXEC (SETREGSET only). + +* The effects of changing the vector length and/or flags are equivalent to + those documented for PR_SME_SET_VL. + + The caller must make a further GETREGSET call if it needs to know what VL is + actually set by SETREGSET, unless is it known in advance that the requested + VL is supported. + +* The size and layout of the payload depends on the header fields. The + SME_PT_ZA_*() macros are provided to facilitate access to the data. + +* In either case, for SETREGSET it is permissible to omit the payload, in which + case the vector length and flags are changed and PSTATE.ZA is set to 0 + (along with any consequences of those changes). If a payload is provided + then PSTATE.ZA will be set to 1. + +* For SETREGSET, if the requested VL is not supported, the effect will be the + same as if the payload were omitted, except that an EIO error is reported. + No attempt is made to translate the payload data to the correct layout + for the vector length actually set. It is up to the caller to translate the + payload layout for the actual VL and retry. + +* The effect of writing a partial, incomplete payload is unspecified. + + +8. ELF coredump extensions +--------------------------- + +* NT_ARM_SSVE notes will be added to each coredump for + each thread of the dumped process. The contents will be equivalent to the + data that would have been read if a PTRACE_GETREGSET of the corresponding + type were executed for each thread when the coredump was generated. + +* A NT_ARM_ZA note will be added to each coredump for each thread of the + dumped process. The contents will be equivalent to the data that would have + been read if a PTRACE_GETREGSET of NT_ARM_ZA were executed for each thread + when the coredump was generated. + + +9. System runtime configuration +-------------------------------- + +* To mitigate the ABI impact of expansion of the signal frame, a policy + mechanism is provided for administrators, distro maintainers and developers + to set the default vector length for userspace processes: + +/proc/sys/abi/sme_default_vector_length + + Writing the text representation of an integer to this file sets the system + default vector length to the specified value, unless the value is greater + than the maximum vector length supported by the system in which case the + default vector length is set to that maximum. + + The result can be determined by reopening the file and reading its + contents. + + At boot, the default vector length is initially set to 32 or the maximum + supported vector length, whichever is smaller and supported. This + determines the initial vector length of the init process (PID 1). + + Reading this file returns the current system default vector length. + +* At every execve() call, the new vector length of the new process is set to + the system default vector length, unless + + * PR_SME_VL_INHERIT (or equivalently SME_PT_VL_INHERIT) is set for the + calling thread, or + + * a deferred vector length change is pending, established via the + PR_SME_SET_VL_ONEXEC flag (or SME_PT_VL_ONEXEC). + +* Modifying the system default vector length does not affect the vector length + of any existing process or thread that does not make an execve() call. + + +Appendix A. SME programmer's model (informative) +================================================= + +This section provides a minimal description of the additions made by SVE to the +ARMv8-A programmer's model that are relevant to this document. + +Note: This section is for information only and not intended to be complete or +to replace any architectural specification. + +A.1. Registers +--------------- + +In A64 state, SME adds the following: + +* A new mode, streaming mode, in which a subset of the normal FPSIMD and SVE + features are available. When supported EL0 software may enter and leave + streaming mode at any time. + + For best system performance it is strongly encouraged for software to enable + streaming mode only when it is actively being used. + +* A new vector length controlling the size of ZA and the Z registers when in + streaming mode, separately to the vector length used for SVE when not in + streaming mode. There is no requirement that either the currently selected + vector length or the set of vector lengths supported for the two modes in + a given system have any relationship. The streaming mode vector length + is referred to as SVL. + +* A new ZA matrix register. This is a square matrix of SVLxSVL bits. Most + operations on ZA require that streaming mode be enabled but ZA can be + enabled without streaming mode in order to load, save and retain data. + + For best system performance it is strongly encouraged for software to enable + ZA only when it is actively being used. + +* Two new 1 bit fields in PSTATE which may be controlled via the SMSTART and + SMSTOP instructions or by access to the SVCR system register: + + * PSTATE.ZA, if this is 1 then the ZA matrix is accessible and has valid + data while if it is 0 then ZA can not be accessed. When PSTATE.ZA is + changed from 0 to 1 all bits in ZA are cleared. + + * PSTATE.SM, if this is 1 then the PE is in streaming mode. When the value + of PSTATE.SM is changed then it is implementation defined if the subset + of the floating point register bits valid in both modes may be retained. + Any other bits will be cleared. + + +References +========== + +[1] arch/arm64/include/uapi/asm/sigcontext.h + AArch64 Linux signal ABI definitions + +[2] arch/arm64/include/uapi/asm/ptrace.h + AArch64 Linux ptrace ABI definitions + +[3] Documentation/arm64/cpu-feature-registers.rst diff --git a/Documentation/arm64/sve.rst b/Documentation/arm64/sve.rst index 9d9a4de5bc34..93c2c2990584 100644 --- a/Documentation/arm64/sve.rst +++ b/Documentation/arm64/sve.rst @@ -7,7 +7,9 @@ Author: Dave Martin Dave.Martin@arm.com Date: 4 August 2017

This document outlines briefly the interface provided to userspace by Linux in -order to support use of the ARM Scalable Vector Extension (SVE). +order to support use of the ARM Scalable Vector Extension (SVE), including +interactions with Streaming SVE mode added by the Scalable Matrix Extension +(SME).

This is an outline of the most important features and issues only and not intended to be exhaustive. @@ -23,6 +25,10 @@ model features for SVE is included in Appendix A. * SVE registers Z0..Z31, P0..P15 and FFR and the current vector length VL, are tracked per-thread.

+* In streaming mode FFR is not accessible unless HWCAP2_SME_FA64 is present + in the system, when it is not supported and these interfaces are used to + access streaming mode FFR is read and written as zero. + * The presence of SVE is reported to userspace via HWCAP_SVE in the aux vector AT_HWCAP entry. Presence of this flag implies the presence of the SVE instructions and registers, and the Linux-specific system interfaces @@ -53,10 +59,19 @@ model features for SVE is included in Appendix A. which userspace can read using an MRS instruction. See elf_hwcaps.txt and cpu-feature-registers.txt for details.

+* On hardware that supports the SME extensions, HWCAP2_SME will also be + reported in the AT_HWCAP2 aux vector entry. Among other things SME adds + streaming mode which provides a subset of the SVE feature set using a + separate SME vector length and the same Z/V registers. See sme.rst + for more details. + * Debuggers should restrict themselves to interacting with the target via the NT_ARM_SVE regset. The recommended way of detecting support for this regset is to connect to a target process first and then attempt a - ptrace(PTRACE_GETREGSET, pid, NT_ARM_SVE, &iov). + ptrace(PTRACE_GETREGSET, pid, NT_ARM_SVE, &iov). Note that when SME is + present and streaming SVE mode is in use the FPSIMD subset of registers + will be read via NT_ARM_SVE and NT_ARM_SVE writes will exit streaming mode + in the target.

* Whenever SVE scalable register values (Zn, Pn, FFR) are exchanged in memory between userspace and the kernel, the register value is encoded in memory in @@ -126,6 +141,11 @@ the SVE instruction set architecture. are only present in fpsimd_context. For convenience, the content of V0..V31 is duplicated between sve_context and fpsimd_context.

+* The record contains a flag field which includes a flag SVE_SIG_FLAG_SM which + if set indicates that the thread is in streaming mode and the vector length + and register data (if present) describe the streaming SVE data and vector + length. + * The signal frame record for SVE always contains basic metadata, in particular the thread's vector length (in sve_context.vl).

@@ -170,6 +190,11 @@ When returning from a signal handler: the signal frame does not match the current vector length, the signal return attempt is treated as illegal, resulting in a forced SIGSEGV.

+* It is permitted to enter or leave streaming mode by setting or clearing + the SVE_SIG_FLAG_SM flag but applications should take care to ensure that + when doing so sve_context.vl and any register data are appropriate for the + vector length in the new mode. +

6. prctl extensions -------------------- @@ -265,8 +290,14 @@ prctl(PR_SVE_GET_VL) 7. ptrace extensions ---------------------

-* A new regset NT_ARM_SVE is defined for use with PTRACE_GETREGSET and - PTRACE_SETREGSET. +* New regsets NT_ARM_SVE and NT_ARM_SSVE are defined for use with + PTRACE_GETREGSET and PTRACE_SETREGSET. NT_ARM_SSVE describes the + streaming mode SVE registers and NT_ARM_SVE describes the + non-streaming mode SVE registers. + + In this description a register set is referred to as being "live" when + the target is in the appropriate streaming or non-streaming mode and is + using data beyond the subset shared with the FPSIMD Vn registers.

Refer to [2] for definitions.

@@ -297,7 +328,7 @@ The regset data starts with struct user_sve_header, containing:

flags

- either + at most one of

SVE_PT_REGS_FPSIMD

@@ -331,6 +362,10 @@ The regset data starts with struct user_sve_header, containing:

SVE_PT_VL_ONEXEC (SETREGSET only).

+ If neither FPSIMD nor SVE flags are provided then no register + payload is available, this is only possible when SME is implemented. + + * The effects of changing the vector length and/or flags are equivalent to those documented for PR_SVE_SET_VL.

@@ -346,6 +381,13 @@ The regset data starts with struct user_sve_header, containing: case only the vector length and flags are changed (along with any consequences of those changes).

+* In systems supporting SME when in streaming mode a GETREGSET for + NT_REG_SVE will return only the user_sve_header with no register data, + similarly a GETREGSET for NT_REG_SSVE will not return any register data + when not in streaming mode. + +* A GETREGSET for NT_ARM_SSVE will never return SVE_PT_REGS_FPSIMD. + * For SETREGSET, if an SVE_PT_REGS_SVE payload is present and the requested VL is not supported, the effect will be the same as if the payload were omitted, except that an EIO error is reported. No @@ -355,17 +397,25 @@ The regset data starts with struct user_sve_header, containing: unspecified. It is up to the caller to translate the payload layout for the actual VL and retry.

+* Where SME is implemented it is not possible to GETREGSET the register + state for normal SVE when in streaming mode, nor the streaming mode + register state when in normal mode, regardless of the implementation defined + behaviour of the hardware for sharing data between the two modes. + +* Any SETREGSET of NT_ARM_SVE will exit streaming mode if the target was in + streaming mode and any SETREGSET of NT_ARM_SSVE will enter streaming mode + if the target was not in streaming mode. + * The effect of writing a partial, incomplete payload is unspecified.

8. ELF coredump extensions ---------------------------

-* A NT_ARM_SVE note will be added to each coredump for each thread of the - dumped process. The contents will be equivalent to the data that would have - been read if a PTRACE_GETREGSET of NT_ARM_SVE were executed for each thread - when the coredump was generated. - +* NT_ARM_SVE and NT_ARM_SSVE notes will be added to each coredump for + each thread of the dumped process. The contents will be equivalent to the + data that would have been read if a PTRACE_GETREGSET of the corresponding + type were executed for each thread when the coredump was generated.

9. System runtime configuration --------------------------------

The arm64 Scalable Matrix Extension (SME) adds some new system registers, fields in existing system registers and exception syndromes. This patch adds definitions for these for use in future patches implementing support for this extension.

Since SME will be the first user of FEAT_HCX in the kernel also include the definitions for enumerating it and the HCRX system register it adds.

Signed-off-by: Mark Brown broonie@kernel.org Acked-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/include/asm/esr.h | 12 +++++- arch/arm64/include/asm/kvm_arm.h | 1 + arch/arm64/include/asm/sysreg.h | 67 ++++++++++++++++++++++++++++++++ arch/arm64/kernel/traps.c | 1 + 4 files changed, 80 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/include/asm/esr.h b/arch/arm64/include/asm/esr.h index d52a0b269ee8..43872e0cfd1e 100644 --- a/arch/arm64/include/asm/esr.h +++ b/arch/arm64/include/asm/esr.h @@ -37,7 +37,8 @@ #define ESR_ELx_EC_ERET (0x1a) /* EL2 only */ /* Unallocated EC: 0x1B */ #define ESR_ELx_EC_FPAC (0x1C) /* EL1 and above */ -/* Unallocated EC: 0x1D - 0x1E */ +#define ESR_ELx_EC_SME (0x1D) +/* Unallocated EC: 0x1E */ #define ESR_ELx_EC_IMP_DEF (0x1f) /* EL3 only */ #define ESR_ELx_EC_IABT_LOW (0x20) #define ESR_ELx_EC_IABT_CUR (0x21) @@ -327,6 +328,15 @@ #define ESR_ELx_CP15_32_ISS_SYS_CNTFRQ (ESR_ELx_CP15_32_ISS_SYS_VAL(0, 0, 14, 0) |\ ESR_ELx_CP15_32_ISS_DIR_READ)

+/* + * ISS values for SME traps + */ + +#define ESR_ELx_SME_ISS_SME_DISABLED 0 +#define ESR_ELx_SME_ISS_ILL 1 +#define ESR_ELx_SME_ISS_SM_DISABLED 2 +#define ESR_ELx_SME_ISS_ZA_DISABLED 3 + #ifndef __ASSEMBLY__ #include <asm/types.h>

diff --git a/arch/arm64/include/asm/kvm_arm.h b/arch/arm64/include/asm/kvm_arm.h index 1767ded83888..13ae232ec4a1 100644 --- a/arch/arm64/include/asm/kvm_arm.h +++ b/arch/arm64/include/asm/kvm_arm.h @@ -279,6 +279,7 @@ #define CPTR_EL2_TCPAC (1U << 31) #define CPTR_EL2_TAM (1 << 30) #define CPTR_EL2_TTA (1 << 20) +#define CPTR_EL2_TSM (1 << 12) #define CPTR_EL2_TFP (1 << CPTR_EL2_TFP_SHIFT) #define CPTR_EL2_TZ (1 << 8) #define CPTR_NVHE_EL2_RES1 0x000032ff /* known RES1 bits in CPTR_EL2 (nVHE) */ diff --git a/arch/arm64/include/asm/sysreg.h b/arch/arm64/include/asm/sysreg.h index fbf5f8bb9055..bebfdd27296a 100644 --- a/arch/arm64/include/asm/sysreg.h +++ b/arch/arm64/include/asm/sysreg.h @@ -118,6 +118,10 @@ * System registers, organised loosely by encoding but grouped together * where the architected name contains an index. e.g. ID_MMFR<n>_EL1. */ +#define SYS_SVCR_SMSTOP_SM_EL0 sys_reg(0, 3, 4, 2, 3) +#define SYS_SVCR_SMSTART_SM_EL0 sys_reg(0, 3, 4, 3, 3) +#define SYS_SVCR_SMSTOP_SMZA_EL0 sys_reg(0, 3, 4, 6, 3) + #define SYS_OSDTRRX_EL1 sys_reg(2, 0, 0, 0, 2) #define SYS_MDCCINT_EL1 sys_reg(2, 0, 0, 2, 0) #define SYS_MDSCR_EL1 sys_reg(2, 0, 0, 2, 2) @@ -181,6 +185,7 @@ #define SYS_ID_AA64PFR0_EL1 sys_reg(3, 0, 0, 4, 0) #define SYS_ID_AA64PFR1_EL1 sys_reg(3, 0, 0, 4, 1) #define SYS_ID_AA64ZFR0_EL1 sys_reg(3, 0, 0, 4, 4) +#define SYS_ID_AA64SMFR0_EL1 sys_reg(3, 0, 0, 4, 5)

#define SYS_ID_AA64DFR0_EL1 sys_reg(3, 0, 0, 5, 0) #define SYS_ID_AA64DFR1_EL1 sys_reg(3, 0, 0, 5, 1) @@ -204,6 +209,8 @@

#define SYS_ZCR_EL1 sys_reg(3, 0, 1, 2, 0) #define SYS_TRFCR_EL1 sys_reg(3, 0, 1, 2, 1) +#define SYS_SMPRI_EL1 sys_reg(3, 0, 1, 2, 4) +#define SYS_SMCR_EL1 sys_reg(3, 0, 1, 2, 6)

#define SYS_TTBR0_EL1 sys_reg(3, 0, 2, 0, 0) #define SYS_TTBR1_EL1 sys_reg(3, 0, 2, 0, 1) @@ -396,6 +403,8 @@ #define TRBIDR_ALIGN_MASK GENMASK(3, 0) #define TRBIDR_ALIGN_SHIFT 0

+#define SMPRI_EL1_PRIORITY_MASK 0xf + #define SYS_PMINTENSET_EL1 sys_reg(3, 0, 9, 14, 1) #define SYS_PMINTENCLR_EL1 sys_reg(3, 0, 9, 14, 2)

@@ -451,8 +460,13 @@ #define SYS_CCSIDR_EL1 sys_reg(3, 1, 0, 0, 0) #define SYS_CLIDR_EL1 sys_reg(3, 1, 0, 0, 1) #define SYS_GMID_EL1 sys_reg(3, 1, 0, 0, 4) +#define SYS_SMIDR_EL1 sys_reg(3, 1, 0, 0, 6) #define SYS_AIDR_EL1 sys_reg(3, 1, 0, 0, 7)

+#define SYS_SMIDR_EL1_IMPLEMENTER_SHIFT 24 +#define SYS_SMIDR_EL1_SMPS_SHIFT 15 +#define SYS_SMIDR_EL1_AFFINITY_SHIFT 0 + #define SYS_CSSELR_EL1 sys_reg(3, 2, 0, 0, 0)

#define SYS_CTR_EL0 sys_reg(3, 3, 0, 0, 1) @@ -461,6 +475,10 @@ #define SYS_RNDR_EL0 sys_reg(3, 3, 2, 4, 0) #define SYS_RNDRRS_EL0 sys_reg(3, 3, 2, 4, 1)

+#define SYS_SVCR_EL0 sys_reg(3, 3, 4, 2, 2) +#define SYS_SVCR_EL0_ZA_MASK 2 +#define SYS_SVCR_EL0_SM_MASK 1 + #define SYS_PMCR_EL0 sys_reg(3, 3, 9, 12, 0) #define SYS_PMCNTENSET_EL0 sys_reg(3, 3, 9, 12, 1) #define SYS_PMCNTENCLR_EL0 sys_reg(3, 3, 9, 12, 2) @@ -477,6 +495,7 @@

#define SYS_TPIDR_EL0 sys_reg(3, 3, 13, 0, 2) #define SYS_TPIDRRO_EL0 sys_reg(3, 3, 13, 0, 3) +#define SYS_TPIDR2_EL0 sys_reg(3, 3, 13, 0, 5)

#define SYS_SCXTNUM_EL0 sys_reg(3, 3, 13, 0, 7)

@@ -546,6 +565,9 @@ #define SYS_HFGITR_EL2 sys_reg(3, 4, 1, 1, 6) #define SYS_ZCR_EL2 sys_reg(3, 4, 1, 2, 0) #define SYS_TRFCR_EL2 sys_reg(3, 4, 1, 2, 1) +#define SYS_HCRX_EL2 sys_reg(3, 4, 1, 2, 2) +#define SYS_SMPRIMAP_EL2 sys_reg(3, 4, 1, 2, 5) +#define SYS_SMCR_EL2 sys_reg(3, 4, 1, 2, 6) #define SYS_DACR32_EL2 sys_reg(3, 4, 3, 0, 0) #define SYS_HDFGRTR_EL2 sys_reg(3, 4, 3, 1, 4) #define SYS_HDFGWTR_EL2 sys_reg(3, 4, 3, 1, 5) @@ -605,6 +627,7 @@ #define SYS_SCTLR_EL12 sys_reg(3, 5, 1, 0, 0) #define SYS_CPACR_EL12 sys_reg(3, 5, 1, 0, 2) #define SYS_ZCR_EL12 sys_reg(3, 5, 1, 2, 0) +#define SYS_SMCR_EL12 sys_reg(3, 5, 1, 2, 6) #define SYS_TTBR0_EL12 sys_reg(3, 5, 2, 0, 0) #define SYS_TTBR1_EL12 sys_reg(3, 5, 2, 0, 1) #define SYS_TCR_EL12 sys_reg(3, 5, 2, 0, 2) @@ -628,6 +651,7 @@ #define SYS_CNTV_CVAL_EL02 sys_reg(3, 5, 14, 3, 2)

/* Common SCTLR_ELx flags. */ +#define SCTLR_ELx_ENTP2 (BIT(60)) #define SCTLR_ELx_DSSBS (BIT(44)) #define SCTLR_ELx_ATA (BIT(43))

@@ -836,6 +860,7 @@ #define ID_AA64PFR0_ELx_32BIT_64BIT 0x2

/* id_aa64pfr1 */ +#define ID_AA64PFR1_SME_SHIFT 24 #define ID_AA64PFR1_MPAMFRAC_SHIFT 16 #define ID_AA64PFR1_RASFRAC_SHIFT 12 #define ID_AA64PFR1_MTE_SHIFT 8 @@ -846,6 +871,7 @@ #define ID_AA64PFR1_SSBS_PSTATE_ONLY 1 #define ID_AA64PFR1_SSBS_PSTATE_INSNS 2 #define ID_AA64PFR1_BT_BTI 0x1 +#define ID_AA64PFR1_SME 1

#define ID_AA64PFR1_MTE_NI 0x0 #define ID_AA64PFR1_MTE_EL0 0x1 @@ -874,6 +900,23 @@ #define ID_AA64ZFR0_AES_PMULL 0x2 #define ID_AA64ZFR0_SVEVER_SVE2 0x1

+/* id_aa64smfr0 */ +#define ID_AA64SMFR0_FA64_SHIFT 63 +#define ID_AA64SMFR0_I16I64_SHIFT 52 +#define ID_AA64SMFR0_F64F64_SHIFT 48 +#define ID_AA64SMFR0_I8I32_SHIFT 36 +#define ID_AA64SMFR0_F16F32_SHIFT 35 +#define ID_AA64SMFR0_B16F32_SHIFT 34 +#define ID_AA64SMFR0_F32F32_SHIFT 32 + +#define ID_AA64SMFR0_FA64 0x1 +#define ID_AA64SMFR0_I16I64 0x4 +#define ID_AA64SMFR0_F64F64 0x1 +#define ID_AA64SMFR0_I8I32 0x4 +#define ID_AA64SMFR0_F16F32 0x1 +#define ID_AA64SMFR0_B16F32 0x1 +#define ID_AA64SMFR0_F32F32 0x1 + /* id_aa64mmfr0 */ #define ID_AA64MMFR0_ECV_SHIFT 60 #define ID_AA64MMFR0_FGT_SHIFT 56 @@ -926,6 +969,7 @@

/* id_aa64mmfr1 */ #define ID_AA64MMFR1_ECBHB_SHIFT 60 +#define ID_AA64MMFR1_HCX_SHIFT 40 #define ID_AA64MMFR1_AFP_SHIFT 44 #define ID_AA64MMFR1_ETS_SHIFT 36 #define ID_AA64MMFR1_TWED_SHIFT 32 @@ -1119,9 +1163,24 @@ #define ZCR_ELx_LEN_SIZE 9 #define ZCR_ELx_LEN_MASK 0x1ff

+#define SMCR_ELx_FA64_SHIFT 31 +#define SMCR_ELx_FA64_MASK (1 << SMCR_ELx_FA64_SHIFT) + +/* + * The SMCR_ELx_LEN_* definitions intentionally include bits [8:4] which + * are reserved by the SME architecture for future expansion of the LEN + * field, with compatible semantics. + */ +#define SMCR_ELx_LEN_SHIFT 0 +#define SMCR_ELx_LEN_SIZE 9 +#define SMCR_ELx_LEN_MASK 0x1ff + #define CPACR_EL1_FPEN_EL1EN (BIT(20)) /* enable EL1 access */ #define CPACR_EL1_FPEN_EL0EN (BIT(21)) /* enable EL0 access, if EL1EN set */

+#define CPACR_EL1_SMEN_EL1EN (BIT(24)) /* enable EL1 access */ +#define CPACR_EL1_SMEN_EL0EN (BIT(25)) /* enable EL0 access, if EL1EN set */ + #define CPACR_EL1_ZEN_EL1EN (BIT(16)) /* enable EL1 access */ #define CPACR_EL1_ZEN_EL0EN (BIT(17)) /* enable EL0 access, if EL1EN set */

@@ -1170,6 +1229,8 @@ #define TRFCR_ELx_ExTRE BIT(1) #define TRFCR_ELx_E0TRE BIT(0)

+/* HCRX_EL2 definitions */ +#define HCRX_EL2_SMPME_MASK (1 << 5)

/* GIC Hypervisor interface registers */ /* ICH_MISR_EL2 bit definitions */ @@ -1233,6 +1294,12 @@ #define ICH_VTR_TDS_SHIFT 19 #define ICH_VTR_TDS_MASK (1 << ICH_VTR_TDS_SHIFT)

+/* HFG[WR]TR_EL2 bit definitions */ +#define HFGxTR_EL2_nTPIDR2_EL0_SHIFT 55 +#define HFGxTR_EL2_nTPIDR2_EL0_MASK BIT_MASK(HFGxTR_EL2_nTPIDR2_EL0_SHIFT) +#define HFGxTR_EL2_nSMPRI_EL1_SHIFT 54 +#define HFGxTR_EL2_nSMPRI_EL1_MASK BIT_MASK(HFGxTR_EL2_nSMPRI_EL1_SHIFT) + #define ARM64_FEATURE_FIELD_BITS 4

/* Create a mask for the feature bits of the specified feature. */ diff --git a/arch/arm64/kernel/traps.c b/arch/arm64/kernel/traps.c index 0529fd57567e..6751621e5bea 100644 --- a/arch/arm64/kernel/traps.c +++ b/arch/arm64/kernel/traps.c @@ -821,6 +821,7 @@ static const char *esr_class_str[] = { [ESR_ELx_EC_SVE] = "SVE", [ESR_ELx_EC_ERET] = "ERET/ERETAA/ERETAB", [ESR_ELx_EC_FPAC] = "FPAC", + [ESR_ELx_EC_SME] = "SME", [ESR_ELx_EC_IMP_DEF] = "EL3 IMP DEF", [ESR_ELx_EC_IABT_LOW] = "IABT (lower EL)", [ESR_ELx_EC_IABT_CUR] = "IABT (current EL)",

SME requires similar setup to that for SVE: disable traps to EL2 and make sure that the maximum vector length is available to EL1, for SME we have two traps - one for SME itself and one for TPIDR2.

In addition since we currently make no active use of priority control for SCMUs we map all SME priorities lower ELs may configure to 0, the architecture specified minimum priority, to ensure that nothing we manage is able to configure itself to consume excessive resources. This will need to be revisited should there be a need to manage SME priorities at runtime.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/include/asm/el2_setup.h | 64 ++++++++++++++++++++++++++++-- 1 file changed, 60 insertions(+), 4 deletions(-)

diff --git a/arch/arm64/include/asm/el2_setup.h b/arch/arm64/include/asm/el2_setup.h index 7f3c87f7a0ce..6430eac98c58 100644 --- a/arch/arm64/include/asm/el2_setup.h +++ b/arch/arm64/include/asm/el2_setup.h @@ -143,6 +143,50 @@ .Lskip_sve_@: .endm

+/* SME register access and priority mapping */ +.macro __init_el2_nvhe_sme + mrs x1, id_aa64pfr1_el1 + ubfx x1, x1, #ID_AA64PFR1_SME_SHIFT, #4 + cbz x1, .Lskip_sme_@ + + bic x0, x0, #CPTR_EL2_TSM // Also disable SME traps + msr cptr_el2, x0 // Disable copro. traps to EL2 + isb + + mrs x1, sctlr_el2 + orr x1, x1, #SCTLR_ELx_ENTP2 // Disable TPIDR2 traps + msr sctlr_el2, x1 + isb + + mov x1, #0 // SMCR controls + + mrs_s x2, SYS_ID_AA64SMFR0_EL1 + ubfx x2, x2, #ID_AA64SMFR0_FA64_SHIFT, #1 // Full FP in SM? + cbz x2, .Lskip_sme_fa64_@ + + orr x1, x1, SMCR_ELx_FA64_MASK +.Lskip_sme_fa64_@: + + orr x1, x1, #SMCR_ELx_LEN_MASK // Enable full SME vector + msr_s SYS_SMCR_EL2, x1 // length for EL1. + + mrs_s x1, SYS_SMIDR_EL1 // Priority mapping supported? + ubfx x1, x1, #SYS_SMIDR_EL1_SMPS_SHIFT, #1 + cbz x1, .Lskip_sme_@ + + msr_s SYS_SMPRIMAP_EL2, xzr // Make all priorities equal + + mrs x1, id_aa64mmfr1_el1 // HCRX_EL2 present? + ubfx x1, x1, #ID_AA64MMFR1_HCX_SHIFT, #4 + cbz x1, .Lskip_sme_@ + + mrs_s x1, SYS_HCRX_EL2 + orr x1, x1, #HCRX_EL2_SMPME_MASK // Enable priority mapping + msr_s SYS_HCRX_EL2, x1 + +.Lskip_sme_@: +.endm + /* Disable any fine grained traps */ .macro __init_el2_fgt mrs x1, id_aa64mmfr0_el1 @@ -153,15 +197,26 @@ mrs x1, id_aa64dfr0_el1 ubfx x1, x1, #ID_AA64DFR0_PMSVER_SHIFT, #4 cmp x1, #3 - b.lt .Lset_fgt_@ + b.lt .Lset_debug_fgt_@ /* Disable PMSNEVFR_EL1 read and write traps */ orr x0, x0, #(1 << 62)

-.Lset_fgt_@: +.Lset_debug_fgt_@: msr_s SYS_HDFGRTR_EL2, x0 msr_s SYS_HDFGWTR_EL2, x0 - msr_s SYS_HFGRTR_EL2, xzr - msr_s SYS_HFGWTR_EL2, xzr + + mov x0, xzr + mrs x1, id_aa64pfr1_el1 + ubfx x1, x1, #ID_AA64PFR1_SME_SHIFT, #4 + cbz x1, .Lset_fgt_@ + + /* Disable nVHE traps of TPIDR2 and SMPRI */ + orr x0, x0, #HFGxTR_EL2_nSMPRI_EL1_MASK + orr x0, x0, #HFGxTR_EL2_nTPIDR2_EL0_MASK + +.Lset_fgt_@: + msr_s SYS_HFGRTR_EL2, x0 + msr_s SYS_HFGWTR_EL2, x0 msr_s SYS_HFGITR_EL2, xzr

mrs x1, id_aa64pfr0_el1 // AMU traps UNDEF without AMU @@ -196,6 +251,7 @@ __init_el2_nvhe_idregs __init_el2_nvhe_cptr __init_el2_nvhe_sve + __init_el2_nvhe_sme __init_el2_fgt __init_el2_nvhe_prepare_eret .endm

The vector lengths used for SME are controlled through a similar set of registers to those for SVE and enumerated using a similar algorithm with some slight differences due to the fact that unlike SVE there are no restrictions on which combinations of vector lengths can be supported nor any mandatory vector lengths which must be implemented. Add a new vector type and implement support for enumerating it.

One slightly awkward feature is that we need to read the current vector length using a different instruction (or enter streaming mode which would have the same issue and be higher cost). Rather than add an ops structure we add special cases directly in the otherwise generic vec_probe_vqs() function, this is a bit inelegant but it's the only place where this is an issue.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/include/asm/cpu.h | 3 + arch/arm64/include/asm/cpufeature.h | 7 ++ arch/arm64/include/asm/fpsimd.h | 26 ++++++ arch/arm64/include/asm/processor.h | 1 + arch/arm64/kernel/cpufeature.c | 47 +++++++++++ arch/arm64/kernel/cpuinfo.c | 4 + arch/arm64/kernel/entry-fpsimd.S | 9 ++ arch/arm64/kernel/fpsimd.c | 123 +++++++++++++++++++++++++++- 8 files changed, 218 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/include/asm/cpu.h b/arch/arm64/include/asm/cpu.h index d08062bcb9c1..115cdec1ae87 100644 --- a/arch/arm64/include/asm/cpu.h +++ b/arch/arm64/include/asm/cpu.h @@ -64,6 +64,9 @@ struct cpuinfo_arm64 {

/* pseudo-ZCR for recording maximum ZCR_EL1 LEN value: */ u64 reg_zcr; + + /* pseudo-SMCR for recording maximum SMCR_EL1 LEN value: */ + u64 reg_smcr; };

DECLARE_PER_CPU(struct cpuinfo_arm64, cpu_data); diff --git a/arch/arm64/include/asm/cpufeature.h b/arch/arm64/include/asm/cpufeature.h index 8ac12e4094aa..5ddfae233ea5 100644 --- a/arch/arm64/include/asm/cpufeature.h +++ b/arch/arm64/include/asm/cpufeature.h @@ -622,6 +622,13 @@ static inline bool id_aa64pfr0_sve(u64 pfr0) return val > 0; }

+static inline bool id_aa64pfr1_sme(u64 pfr1) +{ + u32 val = cpuid_feature_extract_unsigned_field(pfr1, ID_AA64PFR1_SME_SHIFT); + + return val > 0; +} + static inline bool id_aa64pfr1_mte(u64 pfr1) { u32 val = cpuid_feature_extract_unsigned_field(pfr1, ID_AA64PFR1_MTE_SHIFT); diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h index 2e8ef00e7520..32cd682258d9 100644 --- a/arch/arm64/include/asm/fpsimd.h +++ b/arch/arm64/include/asm/fpsimd.h @@ -78,6 +78,7 @@ extern void sme_kernel_enable(const struct arm64_cpu_capabilities *__unused); extern void fa64_kernel_enable(const struct arm64_cpu_capabilities *__unused);

extern u64 read_zcr_features(void); +extern u64 read_smcr_features(void);

/* * Helpers to translate bit indices in sve_vq_map to VQ values (and @@ -172,6 +173,12 @@ static inline void write_vl(enum vec_type type, u64 val) tmp = read_sysreg_s(SYS_ZCR_EL1) & ~ZCR_ELx_LEN_MASK; write_sysreg_s(tmp | val, SYS_ZCR_EL1); break; +#endif +#ifdef CONFIG_ARM64_SME + case ARM64_VEC_SME: + tmp = read_sysreg_s(SYS_SMCR_EL1) & ~SMCR_ELx_LEN_MASK; + write_sysreg_s(tmp | val, SYS_SMCR_EL1); + break; #endif default: WARN_ON_ONCE(1); @@ -268,12 +275,31 @@ static inline void sme_smstop(void) asm volatile(__msr_s(SYS_SVCR_SMSTOP_SMZA_EL0, "xzr")); }

+extern void __init sme_setup(void); + +static inline int sme_max_vl(void) +{ + return vec_max_vl(ARM64_VEC_SME); +} + +static inline int sme_max_virtualisable_vl(void) +{ + return vec_max_virtualisable_vl(ARM64_VEC_SME); +} + +extern unsigned int sme_get_vl(void); + #else

static inline void sme_smstart_sm(void) { } static inline void sme_smstop_sm(void) { } static inline void sme_smstop(void) { }

+static inline void sme_setup(void) { } +static inline unsigned int sme_get_vl(void) { return 0; } +static inline int sme_max_vl(void) { return 0; } +static inline int sme_max_virtualisable_vl(void) { return 0; } + #endif /* ! CONFIG_ARM64_SME */

/* For use by EFI runtime services calls only */ diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h index 73e38d9a540c..abf34a9c2eab 100644 --- a/arch/arm64/include/asm/processor.h +++ b/arch/arm64/include/asm/processor.h @@ -118,6 +118,7 @@ struct debug_info {

enum vec_type { ARM64_VEC_SVE = 0, + ARM64_VEC_SME, ARM64_VEC_MAX, };

diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c index 0f2d7ddd69ae..082b3f48cbfd 100644 --- a/arch/arm64/kernel/cpufeature.c +++ b/arch/arm64/kernel/cpufeature.c @@ -581,6 +581,12 @@ static const struct arm64_ftr_bits ftr_zcr[] = { ARM64_FTR_END, };

+static const struct arm64_ftr_bits ftr_smcr[] = { + ARM64_FTR_BITS(FTR_HIDDEN, FTR_NONSTRICT, FTR_LOWER_SAFE, + SMCR_ELx_LEN_SHIFT, SMCR_ELx_LEN_SIZE, 0), /* LEN */ + ARM64_FTR_END, +}; + /* * Common ftr bits for a 32bit register with all hidden, strict * attributes, with 4bit feature fields and a default safe value of @@ -687,6 +693,7 @@ static const struct __ftr_reg_entry {

/* Op1 = 0, CRn = 1, CRm = 2 */ ARM64_FTR_REG(SYS_ZCR_EL1, ftr_zcr), + ARM64_FTR_REG(SYS_SMCR_EL1, ftr_smcr),

/* Op1 = 1, CRn = 0, CRm = 0 */ ARM64_FTR_REG(SYS_GMID_EL1, ftr_gmid), @@ -991,6 +998,12 @@ void __init init_cpu_features(struct cpuinfo_arm64 *info) vec_init_vq_map(ARM64_VEC_SVE); }

+ if (id_aa64pfr1_sme(info->reg_id_aa64pfr1)) { + init_cpu_ftr_reg(SYS_SMCR_EL1, info->reg_smcr); + if (IS_ENABLED(CONFIG_ARM64_SME)) + vec_init_vq_map(ARM64_VEC_SME); + } + if (id_aa64pfr1_mte(info->reg_id_aa64pfr1)) init_cpu_ftr_reg(SYS_GMID_EL1, info->reg_gmid);

@@ -1217,6 +1230,9 @@ void update_cpu_features(int cpu, taint |= check_update_ftr_reg(SYS_ID_AA64ZFR0_EL1, cpu, info->reg_id_aa64zfr0, boot->reg_id_aa64zfr0);

+ taint |= check_update_ftr_reg(SYS_ID_AA64SMFR0_EL1, cpu, + info->reg_id_aa64smfr0, boot->reg_id_aa64smfr0); + if (id_aa64pfr0_sve(info->reg_id_aa64pfr0)) { taint |= check_update_ftr_reg(SYS_ZCR_EL1, cpu, info->reg_zcr, boot->reg_zcr); @@ -1227,6 +1243,16 @@ void update_cpu_features(int cpu, vec_update_vq_map(ARM64_VEC_SVE); }

+ if (id_aa64pfr1_sme(info->reg_id_aa64pfr1)) { + taint |= check_update_ftr_reg(SYS_SMCR_EL1, cpu, + info->reg_smcr, boot->reg_smcr); + + /* Probe vector lengths, unless we already gave up on SME */ + if (id_aa64pfr1_sme(read_sanitised_ftr_reg(SYS_ID_AA64PFR1_EL1)) && + !system_capabilities_finalized()) + vec_update_vq_map(ARM64_VEC_SME); + } + /* * The kernel uses the LDGM/STGM instructions and the number of tags * they read/write depends on the GMID_EL1.BS field. Check that the @@ -2931,6 +2957,23 @@ static void verify_sve_features(void) /* Add checks on other ZCR bits here if necessary */ }

+static void verify_sme_features(void) +{ + u64 safe_smcr = read_sanitised_ftr_reg(SYS_SMCR_EL1); + u64 smcr = read_smcr_features(); + + unsigned int safe_len = safe_smcr & SMCR_ELx_LEN_MASK; + unsigned int len = smcr & SMCR_ELx_LEN_MASK; + + if (len < safe_len || vec_verify_vq_map(ARM64_VEC_SME)) { + pr_crit("CPU%d: SME: vector length support mismatch\n", + smp_processor_id()); + cpu_die_early(); + } + + /* Add checks on other SMCR bits here if necessary */ +} + static void verify_hyp_capabilities(void) { u64 safe_mmfr1, mmfr0, mmfr1; @@ -2983,6 +3026,9 @@ static void verify_local_cpu_capabilities(void) if (system_supports_sve()) verify_sve_features();

+ if (system_supports_sme()) + verify_sme_features(); + if (is_hyp_mode_available()) verify_hyp_capabilities(); } @@ -3100,6 +3146,7 @@ void __init setup_cpu_features(void) pr_info("emulated: Privileged Access Never (PAN) using TTBR0_EL1 switching\n");

sve_setup(); + sme_setup(); minsigstksz_setup();

/* Advertise that we have computed the system capabilities */ diff --git a/arch/arm64/kernel/cpuinfo.c b/arch/arm64/kernel/cpuinfo.c index a73fe2888b7e..8a8136a096ac 100644 --- a/arch/arm64/kernel/cpuinfo.c +++ b/arch/arm64/kernel/cpuinfo.c @@ -421,6 +421,10 @@ static void __cpuinfo_store_cpu(struct cpuinfo_arm64 *info) id_aa64pfr0_sve(info->reg_id_aa64pfr0)) info->reg_zcr = read_zcr_features();

+ if (IS_ENABLED(CONFIG_ARM64_SME) && + id_aa64pfr1_sme(info->reg_id_aa64pfr1)) + info->reg_smcr = read_smcr_features(); + cpuinfo_detect_icache_policy(info); }

diff --git a/arch/arm64/kernel/entry-fpsimd.S b/arch/arm64/kernel/entry-fpsimd.S index dc242e269f9a..deee5f01462e 100644 --- a/arch/arm64/kernel/entry-fpsimd.S +++ b/arch/arm64/kernel/entry-fpsimd.S @@ -86,3 +86,12 @@ SYM_FUNC_START(sve_flush_live) SYM_FUNC_END(sve_flush_live)

#endif /* CONFIG_ARM64_SVE */ + +#ifdef CONFIG_ARM64_SME + +SYM_FUNC_START(sme_get_vl) + _sme_rdsvl 0, 1 + ret +SYM_FUNC_END(sme_get_vl) + +#endif /* CONFIG_ARM64_SME */ diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index e4fba0bfb55e..5e5fbd9cba75 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -136,6 +136,12 @@ __ro_after_init struct vl_info vl_info[ARM64_VEC_MAX] = { .max_virtualisable_vl = SVE_VL_MIN, }, #endif +#ifdef CONFIG_ARM64_SME + [ARM64_VEC_SME] = { + .type = ARM64_VEC_SME, + .name = "SME", + }, +#endif };

static unsigned int vec_vl_inherit_flag(enum vec_type type) @@ -186,6 +192,20 @@ extern void __percpu *efi_sve_state;

#endif /* ! CONFIG_ARM64_SVE */

+#ifdef CONFIG_ARM64_SME + +static int get_sme_default_vl(void) +{ + return get_default_vl(ARM64_VEC_SME); +} + +static void set_sme_default_vl(int val) +{ + set_default_vl(ARM64_VEC_SME, val); +} + +#endif + DEFINE_PER_CPU(bool, fpsimd_context_busy); EXPORT_PER_CPU_SYMBOL(fpsimd_context_busy);

@@ -409,6 +429,8 @@ static unsigned int find_supported_vector_length(enum vec_type type,

if (vl > max_vl) vl = max_vl; + if (vl < info->min_vl) + vl = info->min_vl;

bit = find_next_bit(info->vq_map, SVE_VQ_MAX, __vq_to_bit(sve_vq_from_vl(vl))); @@ -770,7 +792,23 @@ static void vec_probe_vqs(struct vl_info *info,

for (vq = SVE_VQ_MAX; vq >= SVE_VQ_MIN; --vq) { write_vl(info->type, vq - 1); /* self-syncing */ - vl = sve_get_vl(); + + switch (info->type) { + case ARM64_VEC_SVE: + vl = sve_get_vl(); + break; + case ARM64_VEC_SME: + vl = sme_get_vl(); + break; + default: + vl = 0; + break; + } + + /* Minimum VL identified? */ + if (sve_vq_from_vl(vl) > vq) + break; + vq = sve_vq_from_vl(vl); /* skip intervening lengths */ set_bit(__vq_to_bit(vq), map); } @@ -1017,7 +1055,88 @@ void fa64_kernel_enable(const struct arm64_cpu_capabilities *__always_unused p) SYS_SMCR_EL1); }

-#endif /* CONFIG_ARM64_SVE */ +/* + * Read the pseudo-SMCR used by cpufeatures to identify the supported + * vector length. + * + * Use only if SME is present. + * This function clobbers the SME vector length. + */ +u64 read_smcr_features(void) +{ + u64 smcr; + unsigned int vq_max; + + sme_kernel_enable(NULL); + sme_smstart_sm(); + + /* + * Set the maximum possible VL. + */ + write_sysreg_s(read_sysreg_s(SYS_SMCR_EL1) | SMCR_ELx_LEN_MASK, + SYS_SMCR_EL1); + + smcr = read_sysreg_s(SYS_SMCR_EL1); + smcr &= ~(u64)SMCR_ELx_LEN_MASK; /* Only the LEN field */ + vq_max = sve_vq_from_vl(sve_get_vl()); + smcr |= vq_max - 1; /* set LEN field to maximum effective value */ + + sme_smstop_sm(); + + return smcr; +} + +void __init sme_setup(void) +{ + struct vl_info *info = &vl_info[ARM64_VEC_SME]; + u64 smcr; + int min_bit; + + if (!system_supports_sme()) + return; + + /* + * SME doesn't require any particular vector length be + * supported but it does require at least one. We should have + * disabled the feature entirely while bringing up CPUs but + * let's double check here. + */ + WARN_ON(bitmap_empty(info->vq_map, SVE_VQ_MAX)); + + min_bit = find_last_bit(info->vq_map, SVE_VQ_MAX); + info->min_vl = sve_vl_from_vq(__bit_to_vq(min_bit)); + + smcr = read_sanitised_ftr_reg(SYS_SMCR_EL1); + info->max_vl = sve_vl_from_vq((smcr & SMCR_ELx_LEN_MASK) + 1); + + /* + * Sanity-check that the max VL we determined through CPU features + * corresponds properly to sme_vq_map. If not, do our best: + */ + if (WARN_ON(info->max_vl != find_supported_vector_length(ARM64_VEC_SME, + info->max_vl))) + info->max_vl = find_supported_vector_length(ARM64_VEC_SME, + info->max_vl); + + WARN_ON(info->min_vl > info->max_vl); + + /* + * For the default VL, pick the maximum supported value <= 32 + * (256 bits) if there is one since this is guaranteed not to + * grow the signal frame when in streaming mode, otherwise the + * minimum available VL will be used. + */ + set_sme_default_vl(find_supported_vector_length(ARM64_VEC_SME, 32)); + + pr_info("SME: minimum available vector length %u bytes per vector\n", + info->min_vl); + pr_info("SME: maximum available vector length %u bytes per vector\n", + info->max_vl); + pr_info("SME: default vector length %u bytes per vector\n", + get_sme_default_vl()); +} + +#endif /* CONFIG_ARM64_SME */

/* * Trapped SVE access

As for SVE provide a prctl() interface which allows processes to configure their SME vector length.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/include/asm/fpsimd.h | 4 ++++ arch/arm64/include/asm/processor.h | 4 +++- arch/arm64/include/asm/thread_info.h | 1 + arch/arm64/kernel/fpsimd.c | 32 ++++++++++++++++++++++++++++ include/uapi/linux/prctl.h | 9 ++++++++ kernel/sys.c | 12 +++++++++++ 6 files changed, 61 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h index 32cd682258d9..38fd6aab7feb 100644 --- a/arch/arm64/include/asm/fpsimd.h +++ b/arch/arm64/include/asm/fpsimd.h @@ -288,6 +288,8 @@ static inline int sme_max_virtualisable_vl(void) }

extern unsigned int sme_get_vl(void); +extern int sme_set_current_vl(unsigned long arg); +extern int sme_get_current_vl(void);

#else

@@ -299,6 +301,8 @@ static inline void sme_setup(void) { } static inline unsigned int sme_get_vl(void) { return 0; } static inline int sme_max_vl(void) { return 0; } static inline int sme_max_virtualisable_vl(void) { return 0; } +static inline int sme_set_current_vl(unsigned long arg) { return -EINVAL; } +static inline int sme_get_current_vl(void) { return -EINVAL; }

#endif /* ! CONFIG_ARM64_SME */

diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h index abf34a9c2eab..7a57cbff8a03 100644 --- a/arch/arm64/include/asm/processor.h +++ b/arch/arm64/include/asm/processor.h @@ -355,9 +355,11 @@ extern void __init minsigstksz_setup(void); */ #include <asm/fpsimd.h>

-/* Userspace interface for PR_SVE_{SET,GET}_VL prctl()s: */ +/* Userspace interface for PR_S[MV]E_{SET,GET}_VL prctl()s: */ #define SVE_SET_VL(arg) sve_set_current_vl(arg) #define SVE_GET_VL() sve_get_current_vl() +#define SME_SET_VL(arg) sme_set_current_vl(arg) +#define SME_GET_VL() sme_get_current_vl()

/* PR_PAC_RESET_KEYS prctl */ #define PAC_RESET_KEYS(tsk, arg) ptrauth_prctl_reset_keys(tsk, arg) diff --git a/arch/arm64/include/asm/thread_info.h b/arch/arm64/include/asm/thread_info.h index e1317b7c4525..4e6b58dcd6f9 100644 --- a/arch/arm64/include/asm/thread_info.h +++ b/arch/arm64/include/asm/thread_info.h @@ -82,6 +82,7 @@ int arch_dup_task_struct(struct task_struct *dst, #define TIF_SVE_VL_INHERIT 24 /* Inherit SVE vl_onexec across exec */ #define TIF_SSBD 25 /* Wants SSB mitigation */ #define TIF_TAGGED_ADDR 26 /* Allow tagged user addresses */ +#define TIF_SME_VL_INHERIT 28 /* Inherit SME vl_onexec across exec */

#define _TIF_SIGPENDING (1 << TIF_SIGPENDING) #define _TIF_NEED_RESCHED (1 << TIF_NEED_RESCHED) diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 754a96563f6f..39f44fcb9b99 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -149,6 +149,8 @@ static unsigned int vec_vl_inherit_flag(enum vec_type type) switch (type) { case ARM64_VEC_SVE: return TIF_SVE_VL_INHERIT; + case ARM64_VEC_SME: + return TIF_SME_VL_INHERIT; default: WARN_ON_ONCE(1); return 0; @@ -807,6 +809,36 @@ int sve_get_current_vl(void) return vec_prctl_status(ARM64_VEC_SVE, 0); }

+#ifdef CONFIG_ARM64_SME +/* PR_SME_SET_VL */ +int sme_set_current_vl(unsigned long arg) +{ + unsigned long vl, flags; + int ret; + + vl = arg & PR_SME_VL_LEN_MASK; + flags = arg & ~vl; + + if (!system_supports_sme() || is_compat_task()) + return -EINVAL; + + ret = vec_set_vector_length(current, ARM64_VEC_SME, vl, flags); + if (ret) + return ret; + + return vec_prctl_status(ARM64_VEC_SME, flags); +} + +/* PR_SME_GET_VL */ +int sme_get_current_vl(void) +{ + if (!system_supports_sme() || is_compat_task()) + return -EINVAL; + + return vec_prctl_status(ARM64_VEC_SME, 0); +} +#endif /* CONFIG_ARM64_SME */ + static void vec_probe_vqs(struct vl_info *info, DECLARE_BITMAP(map, SVE_VQ_MAX)) { diff --git a/include/uapi/linux/prctl.h b/include/uapi/linux/prctl.h index e998764f0262..a5e06dcbba13 100644 --- a/include/uapi/linux/prctl.h +++ b/include/uapi/linux/prctl.h @@ -272,6 +272,15 @@ struct prctl_mm_map { # define PR_SCHED_CORE_SCOPE_THREAD_GROUP 1 # define PR_SCHED_CORE_SCOPE_PROCESS_GROUP 2

+/* arm64 Scalable Matrix Extension controls */ +/* Flag values must be in sync with SVE versions */ +#define PR_SME_SET_VL 63 /* set task vector length */ +# define PR_SME_SET_VL_ONEXEC (1 << 18) /* defer effect until exec */ +#define PR_SME_GET_VL 64 /* get task vector length */ +/* Bits common to PR_SME_SET_VL and PR_SME_GET_VL */ +# define PR_SME_VL_LEN_MASK 0xffff +# define PR_SME_VL_INHERIT (1 << 17) /* inherit across exec */ + #define PR_SET_VMA 0x53564d41 # define PR_SET_VMA_ANON_NAME 0

diff --git a/kernel/sys.c b/kernel/sys.c index 374f83e95239..b911fa6d81ab 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -117,6 +117,12 @@ #ifndef SVE_GET_VL # define SVE_GET_VL() (-EINVAL) #endif +#ifndef SME_SET_VL +# define SME_SET_VL(a) (-EINVAL) +#endif +#ifndef SME_GET_VL +# define SME_GET_VL() (-EINVAL) +#endif #ifndef PAC_RESET_KEYS # define PAC_RESET_KEYS(a, b) (-EINVAL) #endif @@ -2541,6 +2547,12 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3, case PR_SVE_GET_VL: error = SVE_GET_VL(); break; + case PR_SME_SET_VL: + error = SME_SET_VL(arg2); + break; + case PR_SME_GET_VL: + error = SME_GET_VL(); + break; case PR_GET_SPECULATION_CTRL: if (arg3 || arg4 || arg5) return -EINVAL;

In SME the use of both streaming SVE mode and ZA are tracked through PSTATE.SM and PSTATE.ZA, visible through the system register SVCR. In order to context switch the floating point state for SME we need to context switch the contents of this register as part of context switching the floating point state.

Since changing the vector length exits streaming SVE mode and disables ZA we also make sure we update SVCR appropriately when setting vector length, and similarly ensure that new threads have streaming SVE mode and ZA disabled.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/include/asm/fpsimd.h | 3 ++- arch/arm64/include/asm/processor.h | 1 + arch/arm64/include/asm/thread_info.h | 1 + arch/arm64/kernel/fpsimd.c | 18 +++++++++++++++++- arch/arm64/kernel/process.c | 2 ++ arch/arm64/kvm/fpsimd.c | 7 ++++++- 6 files changed, 29 insertions(+), 3 deletions(-)

diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h index 38fd6aab7feb..821d270980da 100644 --- a/arch/arm64/include/asm/fpsimd.h +++ b/arch/arm64/include/asm/fpsimd.h @@ -46,7 +46,8 @@ extern void fpsimd_restore_current_state(void); extern void fpsimd_update_current_state(struct user_fpsimd_state const *state);

extern void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *state, - void *sve_state, unsigned int sve_vl); + void *sve_state, unsigned int sve_vl, + u64 *svcr);

extern void fpsimd_flush_task_state(struct task_struct *target); extern void fpsimd_save_and_flush_cpu_state(void); diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h index 849e97d418a8..22cd11e86854 100644 --- a/arch/arm64/include/asm/processor.h +++ b/arch/arm64/include/asm/processor.h @@ -169,6 +169,7 @@ struct thread_struct { u64 mte_ctrl; #endif u64 sctlr_user; + u64 svcr; u64 tpidr2_el0; };

diff --git a/arch/arm64/include/asm/thread_info.h b/arch/arm64/include/asm/thread_info.h index 4e6b58dcd6f9..848739c15de8 100644 --- a/arch/arm64/include/asm/thread_info.h +++ b/arch/arm64/include/asm/thread_info.h @@ -82,6 +82,7 @@ int arch_dup_task_struct(struct task_struct *dst, #define TIF_SVE_VL_INHERIT 24 /* Inherit SVE vl_onexec across exec */ #define TIF_SSBD 25 /* Wants SSB mitigation */ #define TIF_TAGGED_ADDR 26 /* Allow tagged user addresses */ +#define TIF_SME 27 /* SME in use */ #define TIF_SME_VL_INHERIT 28 /* Inherit SME vl_onexec across exec */

#define _TIF_SIGPENDING (1 << TIF_SIGPENDING) diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 231f2d85b65e..1c113349f6cc 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -121,6 +121,7 @@ struct fpsimd_last_state_struct { struct user_fpsimd_state *st; void *sve_state; + u64 *svcr; unsigned int sve_vl; };

@@ -359,6 +360,9 @@ static void task_fpsimd_load(void) WARN_ON(!system_supports_fpsimd()); WARN_ON(!have_cpu_fpsimd_context());

+ if (IS_ENABLED(CONFIG_ARM64_SME) && test_thread_flag(TIF_SME)) + write_sysreg_s(current->thread.svcr, SYS_SVCR_EL0); + if (IS_ENABLED(CONFIG_ARM64_SVE) && test_thread_flag(TIF_SVE)) { sve_set_vq(sve_vq_from_vl(task_get_sve_vl(current)) - 1); sve_load_state(sve_pffr(&current->thread), @@ -390,6 +394,12 @@ static void fpsimd_save(void) if (test_thread_flag(TIF_FOREIGN_FPSTATE)) return;

+ if (IS_ENABLED(CONFIG_ARM64_SME) && + test_thread_flag(TIF_SME)) { + u64 *svcr = last->svcr; + *svcr = read_sysreg_s(SYS_SVCR_EL0); + } + if (IS_ENABLED(CONFIG_ARM64_SVE) && test_thread_flag(TIF_SVE)) { if (WARN_ON(sve_get_vl() != last->sve_vl)) { @@ -741,6 +751,10 @@ int vec_set_vector_length(struct task_struct *task, enum vec_type type, if (test_and_clear_tsk_thread_flag(task, TIF_SVE)) sve_to_fpsimd(task);

+ if (system_supports_sme() && type == ARM64_VEC_SME) + task->thread.svcr &= ~(SYS_SVCR_EL0_SM_MASK | + SYS_SVCR_EL0_ZA_MASK); + if (task == current) put_cpu_fpsimd_context();

@@ -1404,6 +1418,7 @@ static void fpsimd_bind_task_to_cpu(void) last->st = &current->thread.uw.fpsimd_state; last->sve_state = current->thread.sve_state; last->sve_vl = task_get_sve_vl(current); + last->svcr = &current->thread.svcr; current->thread.fpsimd_cpu = smp_processor_id();

if (system_supports_sve()) { @@ -1418,7 +1433,7 @@ static void fpsimd_bind_task_to_cpu(void) }

void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *st, void *sve_state, - unsigned int sve_vl) + unsigned int sve_vl, u64 *svcr) { struct fpsimd_last_state_struct *last = this_cpu_ptr(&fpsimd_last_state); @@ -1427,6 +1442,7 @@ void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *st, void *sve_state, WARN_ON(!in_softirq() && !irqs_disabled());

last->st = st; + last->svcr = svcr; last->sve_state = sve_state; last->sve_vl = sve_vl; } diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c index e20571f19718..07f235b46cf5 100644 --- a/arch/arm64/kernel/process.c +++ b/arch/arm64/kernel/process.c @@ -310,6 +310,8 @@ int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src) dst->thread.sve_state = NULL; clear_tsk_thread_flag(dst, TIF_SVE);

+ dst->thread.svcr = 0; + /* clear any pending asynchronous tag fault raised by the parent */ clear_tsk_thread_flag(dst, TIF_MTE_ASYNC_FAULT);

diff --git a/arch/arm64/kvm/fpsimd.c b/arch/arm64/kvm/fpsimd.c index 397fdac75cb1..ac09f1f682ff 100644 --- a/arch/arm64/kvm/fpsimd.c +++ b/arch/arm64/kvm/fpsimd.c @@ -109,9 +109,14 @@ void kvm_arch_vcpu_ctxsync_fp(struct kvm_vcpu *vcpu) WARN_ON_ONCE(!irqs_disabled());

if (vcpu->arch.flags & KVM_ARM64_FP_ENABLED) { + /* + * Currently we do not support SME guests so SVCR is + * always 0 and we just need a variable to point to. + */ fpsimd_bind_state_to_cpu(&vcpu->arch.ctxt.fp_regs, vcpu->arch.sve_state, - vcpu->arch.sve_max_vl); + vcpu->arch.sve_max_vl, + NULL);

clear_thread_flag(TIF_FOREIGN_FPSTATE); update_thread_flag(TIF_SVE, vcpu_has_sve(vcpu));

Allocate space for storing ZA on first access to SME and use that to save and restore ZA state when context switching. We do this by using the vector form of the LDR and STR ZA instructions, these do not require streaming mode and have implementation recommendations that they avoid contention issues in shared SMCU implementations.

Since ZA is architecturally guaranteed to be zeroed when enabled we do not need to explicitly zero ZA, either we will be restoring from a saved copy or trapping on first use of SME so we know that ZA must be disabled.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/include/asm/fpsimd.h | 5 ++++- arch/arm64/include/asm/fpsimdmacros.h | 22 ++++++++++++++++++++++ arch/arm64/include/asm/kvm_host.h | 3 +++ arch/arm64/include/asm/processor.h | 1 + arch/arm64/kernel/entry-fpsimd.S | 22 ++++++++++++++++++++++ arch/arm64/kernel/fpsimd.c | 20 +++++++++++++------- arch/arm64/kvm/fpsimd.c | 2 +- 7 files changed, 66 insertions(+), 9 deletions(-)

diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h index cd94f5c5b516..1a709c03bb6c 100644 --- a/arch/arm64/include/asm/fpsimd.h +++ b/arch/arm64/include/asm/fpsimd.h @@ -47,7 +47,8 @@ extern void fpsimd_update_current_state(struct user_fpsimd_state const *state);

extern void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *state, void *sve_state, unsigned int sve_vl, - unsigned int sme_vl, u64 *svcr); + void *za_state, unsigned int sme_vl, + u64 *svcr);

extern void fpsimd_flush_task_state(struct task_struct *target); extern void fpsimd_save_and_flush_cpu_state(void); @@ -90,6 +91,8 @@ extern void sve_flush_live(bool flush_ffr, unsigned long vq_minus_1); extern unsigned int sve_get_vl(void); extern void sve_set_vq(unsigned long vq_minus_1); extern void sme_set_vq(unsigned long vq_minus_1); +extern void za_save_state(void *state); +extern void za_load_state(void const *state);

struct arm64_cpu_capabilities; extern void sve_kernel_enable(const struct arm64_cpu_capabilities *__unused); diff --git a/arch/arm64/include/asm/fpsimdmacros.h b/arch/arm64/include/asm/fpsimdmacros.h index f6ab36e0cd8d..5e0910cf4832 100644 --- a/arch/arm64/include/asm/fpsimdmacros.h +++ b/arch/arm64/include/asm/fpsimdmacros.h @@ -319,3 +319,25 @@ ldr w\nxtmp, [\xpfpsr, #4] msr fpcr, x\nxtmp .endm + +.macro sme_save_za nxbase, xvl, nw + mov w\nw, #0 + +423: + _sme_str_zav \nw, \nxbase + add x\nxbase, x\nxbase, \xvl + add x\nw, x\nw, #1 + cmp \xvl, x\nw + bne 423b +.endm + +.macro sme_load_za nxbase, xvl, nw + mov w\nw, #0 + +423: + _sme_ldr_zav \nw, \nxbase + add x\nxbase, x\nxbase, \xvl + add x\nw, x\nw, #1 + cmp \xvl, x\nw + bne 423b +.endm diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h index e3b25dc6c367..90e92bcf2352 100644 --- a/arch/arm64/include/asm/kvm_host.h +++ b/arch/arm64/include/asm/kvm_host.h @@ -285,8 +285,11 @@ struct vcpu_reset_state {

struct kvm_vcpu_arch { struct kvm_cpu_context ctxt; + + /* Guest floating point state */ void *sve_state; unsigned int sve_max_vl; + u64 svcr;

/* Stage 2 paging state used by the hardware on next switch */ struct kvm_s2_mmu *hw_mmu; diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h index 7542310b4e6b..6a3a6c3dec90 100644 --- a/arch/arm64/include/asm/processor.h +++ b/arch/arm64/include/asm/processor.h @@ -154,6 +154,7 @@ struct thread_struct {

unsigned int fpsimd_cpu; void *sve_state; /* SVE registers, if any */ + void *za_state; /* ZA register, if any */ unsigned int vl[ARM64_VEC_MAX]; /* vector length */ unsigned int vl_onexec[ARM64_VEC_MAX]; /* vl after next exec */ unsigned long fault_address; /* fault info */ diff --git a/arch/arm64/kernel/entry-fpsimd.S b/arch/arm64/kernel/entry-fpsimd.S index 6f88c0f86d50..229436f33df5 100644 --- a/arch/arm64/kernel/entry-fpsimd.S +++ b/arch/arm64/kernel/entry-fpsimd.S @@ -99,4 +99,26 @@ SYM_FUNC_START(sme_set_vq) ret SYM_FUNC_END(sme_set_vq)

+/* + * Save the SME state + * + * x0 - pointer to buffer for state + */ +SYM_FUNC_START(za_save_state) + _sme_rdsvl 1, 1 // x1 = VL/8 + sme_save_za 0, x1, 12 + ret +SYM_FUNC_END(za_save_state) + +/* + * Load the SME state + * + * x0 - pointer to buffer for state + */ +SYM_FUNC_START(za_load_state) + _sme_rdsvl 1, 1 // x1 = VL/8 + sme_load_za 0, x1, 12 + ret +SYM_FUNC_END(za_load_state) + #endif /* CONFIG_ARM64_SME */ diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index f8506a875eb2..dc38f3f2a28a 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -121,6 +121,7 @@ struct fpsimd_last_state_struct { struct user_fpsimd_state *st; void *sve_state; + void *za_state; u64 *svcr; unsigned int sve_vl; unsigned int sme_vl; @@ -387,11 +388,15 @@ static void task_fpsimd_load(void) if (system_supports_sme()) { unsigned long sme_vl = task_get_sme_vl(current);

+ /* Ensure VL is set up for restoring data */ if (test_thread_flag(TIF_SME)) sme_set_vq(sve_vq_from_vl(sme_vl) - 1);

write_sysreg_s(current->thread.svcr, SYS_SVCR_EL0);

+ if (thread_za_enabled(&current->thread)) + za_load_state(current->thread.za_state); + if (thread_sm_enabled(&current->thread)) { restore_sve_regs = true; restore_ffr = system_supports_fa64(); @@ -441,11 +446,10 @@ static void fpsimd_save(void) u64 *svcr = last->svcr; *svcr = read_sysreg_s(SYS_SVCR_EL0);

- if (thread_za_enabled(&current->thread)) { - /* ZA state managment is not implemented yet */ - force_signal_inject(SIGKILL, SI_KERNEL, 0, 0); - return; - } + *svcr = read_sysreg_s(SYS_SVCR_EL0); + + if (*svcr & SYS_SVCR_EL0_ZA_MASK) + za_save_state(last->za_state);

/* If we are in streaming mode override regular SVE. */ if (*svcr & SYS_SVCR_EL0_SM_MASK) { @@ -1483,6 +1487,7 @@ static void fpsimd_bind_task_to_cpu(void) WARN_ON(!system_supports_fpsimd()); last->st = &current->thread.uw.fpsimd_state; last->sve_state = current->thread.sve_state; + last->za_state = current->thread.za_state; last->sve_vl = task_get_sve_vl(current); last->sme_vl = task_get_sme_vl(current); last->svcr = &current->thread.svcr; @@ -1500,8 +1505,8 @@ static void fpsimd_bind_task_to_cpu(void) }

void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *st, void *sve_state, - unsigned int sve_vl, unsigned int sme_vl, - u64 *svcr) + unsigned int sve_vl, void *za_state, + unsigned int sme_vl, u64 *svcr) { struct fpsimd_last_state_struct *last = this_cpu_ptr(&fpsimd_last_state); @@ -1512,6 +1517,7 @@ void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *st, void *sve_state, last->st = st; last->svcr = svcr; last->sve_state = sve_state; + last->za_state = za_state; last->sve_vl = sve_vl; last->sme_vl = sme_vl; } diff --git a/arch/arm64/kvm/fpsimd.c b/arch/arm64/kvm/fpsimd.c index 394e583bb73e..57d7ac3cfa0c 100644 --- a/arch/arm64/kvm/fpsimd.c +++ b/arch/arm64/kvm/fpsimd.c @@ -116,7 +116,7 @@ void kvm_arch_vcpu_ctxsync_fp(struct kvm_vcpu *vcpu) fpsimd_bind_state_to_cpu(&vcpu->arch.ctxt.fp_regs, vcpu->arch.sve_state, vcpu->arch.sve_max_vl, - 0, NULL); + NULL, 0, &vcpu->arch.svcr);

clear_thread_flag(TIF_FOREIGN_FPSTATE); update_thread_flag(TIF_SVE, vcpu_has_sve(vcpu));

The ABI requires that streaming mode and ZA are disabled when invoking signal handlers, do this in setup_return() when we prepare the task state for the signal handler.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/kernel/signal.c | 7 +++++++ 1 file changed, 7 insertions(+)

diff --git a/arch/arm64/kernel/signal.c b/arch/arm64/kernel/signal.c index 4a4122ef6f39..42efa464e46e 100644 --- a/arch/arm64/kernel/signal.c +++ b/arch/arm64/kernel/signal.c @@ -759,6 +759,13 @@ static void setup_return(struct pt_regs *regs, struct k_sigaction *ka, /* TCO (Tag Check Override) always cleared for signal handlers */ regs->pstate &= ~PSR_TCO_BIT;

+ /* Signal handlers are invoked with ZA and streaming mode disabled */ + if (system_supports_sme()) { + current->thread.svcr &= ~(SYS_SVCR_EL0_ZA_MASK | + SYS_SVCR_EL0_SM_MASK); + sme_smstop(); + } + if (ka->sa.sa_flags & SA_RESTORER) sigtramp = ka->sa.sa_restorer; else

Implement support for ZA in signal handling in a very similar way to how we implement support for SVE registers, using a signal context structure with optional register state after it. Where present this register state stores the ZA matrix as a series of horizontal vectors numbered from 0 to VL/8 in the endinanness independent format used for vectors.

As with SVE we do not allow changes in the vector length during signal return but we do allow ZA to be enabled or disabled.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/include/uapi/asm/sigcontext.h | 41 +++++++ arch/arm64/kernel/fpsimd.c | 3 - arch/arm64/kernel/signal.c | 139 +++++++++++++++++++++++ 3 files changed, 180 insertions(+), 3 deletions(-)

diff --git a/arch/arm64/include/uapi/asm/sigcontext.h b/arch/arm64/include/uapi/asm/sigcontext.h index 57e9f8c3ee9e..4aaf31e3bf16 100644 --- a/arch/arm64/include/uapi/asm/sigcontext.h +++ b/arch/arm64/include/uapi/asm/sigcontext.h @@ -140,6 +140,14 @@ struct sve_context {

#define SVE_SIG_FLAG_SM 0x1 /* Context describes streaming mode */

+#define ZA_MAGIC 0x54366345 + +struct za_context { + struct _aarch64_ctx head; + __u16 vl; + __u16 __reserved[3]; +}; + #endif /* !__ASSEMBLY__ */

#include <asm/sve_context.h> @@ -259,4 +267,37 @@ struct sve_context { #define SVE_SIG_CONTEXT_SIZE(vq) \ (SVE_SIG_REGS_OFFSET + SVE_SIG_REGS_SIZE(vq))

+/* + * If the ZA register is enabled for the thread at signal delivery then, + * za_context.head.size >= ZA_SIG_CONTEXT_SIZE(sve_vq_from_vl(za_context.vl)) + * and the register data may be accessed using the ZA_SIG_*() macros. + * + * If za_context.head.size < ZA_SIG_CONTEXT_SIZE(sve_vq_from_vl(za_context.vl)) + * then ZA was not enabled and no register data was included in which case + * ZA register was not enabled for the thread and no register data + * the ZA_SIG_*() macros should not be used except for this check. + * + * The same convention applies when returning from a signal: a caller + * will need to remove or resize the za_context block if it wants to + * enable the ZA register when it was previously non-live or vice-versa. + * This may require the caller to allocate fresh memory and/or move other + * context blocks in the signal frame. + * + * Changing the vector length during signal return is not permitted: + * za_context.vl must equal the thread's current SME vector length when + * doing a sigreturn. + */ + +#define ZA_SIG_REGS_OFFSET \ + ((sizeof(struct za_context) + (__SVE_VQ_BYTES - 1)) \ + / __SVE_VQ_BYTES * __SVE_VQ_BYTES) + +#define ZA_SIG_REGS_SIZE(vq) ((vq * __SVE_VQ_BYTES) * (vq * __SVE_VQ_BYTES)) + +#define ZA_SIG_ZAV_OFFSET(vq, n) (ZA_SIG_REGS_OFFSET + \ + (SVE_SIG_ZREG_SIZE(vq) * n)) + +#define ZA_SIG_CONTEXT_SIZE(vq) \ + (ZA_SIG_REGS_OFFSET + ZA_SIG_REGS_SIZE(vq)) + #endif /* _UAPI__ASM_SIGCONTEXT_H */ diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 00a0cbd01ce5..80f7ca12f855 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -1181,9 +1181,6 @@ void fpsimd_release_task(struct task_struct *dead_task)

#ifdef CONFIG_ARM64_SME

-/* This will move to uapi/asm/sigcontext.h when signals are implemented */ -#define ZA_SIG_REGS_SIZE(vq) ((vq * __SVE_VQ_BYTES) * (vq * __SVE_VQ_BYTES)) - /* * Ensure that task->thread.za_state is allocated and sufficiently large. * diff --git a/arch/arm64/kernel/signal.c b/arch/arm64/kernel/signal.c index 0ddce6afd2a3..2295948d97fd 100644 --- a/arch/arm64/kernel/signal.c +++ b/arch/arm64/kernel/signal.c @@ -56,6 +56,7 @@ struct rt_sigframe_user_layout { unsigned long fpsimd_offset; unsigned long esr_offset; unsigned long sve_offset; + unsigned long za_offset; unsigned long extra_offset; unsigned long end_offset; }; @@ -218,6 +219,7 @@ static int restore_fpsimd_context(struct fpsimd_context __user *ctx) struct user_ctxs { struct fpsimd_context __user *fpsimd; struct sve_context __user *sve; + struct za_context __user *za; };

#ifdef CONFIG_ARM64_SVE @@ -346,6 +348,101 @@ extern int restore_sve_fpsimd_context(struct user_ctxs *user);

#endif /* ! CONFIG_ARM64_SVE */

+#ifdef CONFIG_ARM64_SME + +static int preserve_za_context(struct za_context __user *ctx) +{ + int err = 0; + u16 reserved[ARRAY_SIZE(ctx->__reserved)]; + unsigned int vl = task_get_sme_vl(current); + unsigned int vq; + + if (thread_za_enabled(&current->thread)) + vq = sve_vq_from_vl(vl); + else + vq = 0; + + memset(reserved, 0, sizeof(reserved)); + + __put_user_error(ZA_MAGIC, &ctx->head.magic, err); + __put_user_error(round_up(ZA_SIG_CONTEXT_SIZE(vq), 16), + &ctx->head.size, err); + __put_user_error(vl, &ctx->vl, err); + BUILD_BUG_ON(sizeof(ctx->__reserved) != sizeof(reserved)); + err |= __copy_to_user(&ctx->__reserved, reserved, sizeof(reserved)); + + if (vq) { + /* + * This assumes that the ZA state has already been saved to + * the task struct by calling the function + * fpsimd_signal_preserve_current_state(). + */ + err |= __copy_to_user((char __user *)ctx + ZA_SIG_REGS_OFFSET, + current->thread.za_state, + ZA_SIG_REGS_SIZE(vq)); + } + + return err ? -EFAULT : 0; +} + +static int restore_za_context(struct user_ctxs __user *user) +{ + int err; + unsigned int vq; + struct za_context za; + + if (__copy_from_user(&za, user->za, sizeof(za))) + return -EFAULT; + + if (za.vl != task_get_sme_vl(current)) + return -EINVAL; + + if (za.head.size <= sizeof(*user->za)) { + current->thread.svcr &= ~SYS_SVCR_EL0_ZA_MASK; + return 0; + } + + vq = sve_vq_from_vl(za.vl); + + if (za.head.size < ZA_SIG_CONTEXT_SIZE(vq)) + return -EINVAL; + + /* + * Careful: we are about __copy_from_user() directly into + * thread.za_state with preemption enabled, so protection is + * needed to prevent a racing context switch from writing stale + * registers back over the new data. + */ + + fpsimd_flush_task_state(current); + /* From now, fpsimd_thread_switch() won't touch thread.sve_state */ + + sme_alloc(current); + if (!current->thread.za_state) { + current->thread.svcr &= ~SYS_SVCR_EL0_ZA_MASK; + clear_thread_flag(TIF_SME); + return -ENOMEM; + } + + err = __copy_from_user(current->thread.za_state, + (char __user const *)user->za + + ZA_SIG_REGS_OFFSET, + ZA_SIG_REGS_SIZE(vq)); + if (err) + return -EFAULT; + + set_thread_flag(TIF_SME); + current->thread.svcr |= SYS_SVCR_EL0_ZA_MASK; + + return 0; +} +#else /* ! CONFIG_ARM64_SME */ + +/* Turn any non-optimised out attempts to use these into a link error: */ +extern int preserve_za_context(void __user *ctx); +extern int restore_za_context(struct user_ctxs *user); + +#endif /* ! CONFIG_ARM64_SME */

static int parse_user_sigframe(struct user_ctxs *user, struct rt_sigframe __user *sf) @@ -360,6 +457,7 @@ static int parse_user_sigframe(struct user_ctxs *user,

user->fpsimd = NULL; user->sve = NULL; + user->za = NULL;

if (!IS_ALIGNED((unsigned long)base, 16)) goto invalid; @@ -425,6 +523,19 @@ static int parse_user_sigframe(struct user_ctxs *user, user->sve = (struct sve_context __user *)head; break;

+ case ZA_MAGIC: + if (!system_supports_sme()) + goto invalid; + + if (user->za) + goto invalid; + + if (size < sizeof(*user->za)) + goto invalid; + + user->za = (struct za_context __user *)head; + break; + case EXTRA_MAGIC: if (have_extra_context) goto invalid; @@ -548,6 +659,9 @@ static int restore_sigframe(struct pt_regs *regs, } }

+ if (err == 0 && system_supports_sme() && user.za) + err = restore_za_context(&user); + return err; }

@@ -630,6 +744,24 @@ static int setup_sigframe_layout(struct rt_sigframe_user_layout *user, return err; }

+ if (system_supports_sme()) { + unsigned int vl; + unsigned int vq = 0; + + if (add_all) + vl = sme_max_vl(); + else + vl = task_get_sme_vl(current); + + if (thread_za_enabled(&current->thread)) + vq = sve_vq_from_vl(vl); + + err = sigframe_alloc(user, &user->za_offset, + ZA_SIG_CONTEXT_SIZE(vq)); + if (err) + return err; + } + return sigframe_alloc_end(user); }

@@ -678,6 +810,13 @@ static int setup_sigframe(struct rt_sigframe_user_layout *user, err |= preserve_sve_context(sve_ctx); }

+ /* ZA state if present */ + if (system_supports_sme() && err == 0 && user->za_offset) { + struct za_context __user *za_ctx = + apply_user_offset(user, user->za_offset); + err |= preserve_za_context(za_ctx); + } + if (err == 0 && user->extra_offset) { char __user *sfp = (char __user *)user->sigframe; char __user *userp =

The ZA array can be read and written with the NT_ARM_ZA. Similarly to our interface for the SVE vector registers the regset consists of a header with information on the current vector length followed by an optional register data payload, represented as for signals as a series of horizontal vectors from 0 to VL/8 in the endianness independent format used for vectors.

On get if ZA is enabled then register data will be provided, otherwise it will be omitted. On set if register data is provided then ZA is enabled and initialized using the provided data, otherwise it is disabled.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/include/uapi/asm/ptrace.h | 56 +++++++++++ arch/arm64/kernel/ptrace.c | 144 +++++++++++++++++++++++++++ include/uapi/linux/elf.h | 1 + 3 files changed, 201 insertions(+)

diff --git a/arch/arm64/include/uapi/asm/ptrace.h b/arch/arm64/include/uapi/asm/ptrace.h index 522b925a78c1..7fa2f7036aa7 100644 --- a/arch/arm64/include/uapi/asm/ptrace.h +++ b/arch/arm64/include/uapi/asm/ptrace.h @@ -268,6 +268,62 @@ struct user_pac_generic_keys { __uint128_t apgakey; };

+/* ZA state (NT_ARM_ZA) */ + +struct user_za_header { + __u32 size; /* total meaningful regset content in bytes */ + __u32 max_size; /* maxmium possible size for this thread */ + __u16 vl; /* current vector length */ + __u16 max_vl; /* maximum possible vector length */ + __u16 flags; + __u16 __reserved; +}; + +/* + * Common ZA_PT_* flags: + * These must be kept in sync with prctl interface in <linux/prctl.h> + */ +#define ZA_PT_VL_INHERIT ((1 << 17) /* PR_SME_VL_INHERIT */ >> 16) +#define ZA_PT_VL_ONEXEC ((1 << 18) /* PR_SME_SET_VL_ONEXEC */ >> 16) + + +/* + * The remainder of the ZA state follows struct user_za_header. The + * total size of the ZA state (including header) depends on the + * metadata in the header: ZA_PT_SIZE(vq, flags) gives the total size + * of the state in bytes, including the header. + * + * Refer to <asm/sigcontext.h> for details of how to pass the correct + * "vq" argument to these macros. + */ + +/* Offset from the start of struct user_za_header to the register data */ +#define ZA_PT_ZA_OFFSET \ + ((sizeof(struct user_za_header) + (__SVE_VQ_BYTES - 1)) \ + / __SVE_VQ_BYTES * __SVE_VQ_BYTES) + +/* + * The payload starts at offset ZA_PT_ZA_OFFSET, and is of size + * ZA_PT_ZA_SIZE(vq, flags). + * + * The ZA array is stored as a sequence of horizontal vectors ZAV of SVL/8 + * bytes each, starting from vector 0. + * + * Additional data might be appended in the future. + * + * The ZA matrix is represented in memory in an endianness-invariant layout + * which differs from the layout used for the FPSIMD V-registers on big-endian + * systems: see sigcontext.h for more explanation. + */ + +#define ZA_PT_ZAV_OFFSET(vq, n) \ + (ZA_PT_ZA_OFFSET + ((vq * __SVE_VQ_BYTES) * n)) + +#define ZA_PT_ZA_SIZE(vq) ((vq * __SVE_VQ_BYTES) * (vq * __SVE_VQ_BYTES)) + +#define ZA_PT_SIZE(vq) \ + (ZA_PT_ZA_OFFSET + ZA_PT_ZA_SIZE(vq)) + #endif /* __ASSEMBLY__ */

#endif /* _UAPI__ASM_PTRACE_H */ diff --git a/arch/arm64/kernel/ptrace.c b/arch/arm64/kernel/ptrace.c index 60185c27b394..47d8a7472171 100644 --- a/arch/arm64/kernel/ptrace.c +++ b/arch/arm64/kernel/ptrace.c @@ -997,6 +997,141 @@ static int ssve_set(struct task_struct *target, ARM64_VEC_SME); }

+static int za_get(struct task_struct *target, + const struct user_regset *regset, + struct membuf to) +{ + struct user_za_header header; + unsigned int vq; + unsigned long start, end; + + if (!system_supports_sme()) + return -EINVAL; + + /* Header */ + memset(&header, 0, sizeof(header)); + + if (test_tsk_thread_flag(target, TIF_SME_VL_INHERIT)) + header.flags |= ZA_PT_VL_INHERIT; + + header.vl = task_get_sme_vl(target); + vq = sve_vq_from_vl(header.vl); + header.max_vl = sme_max_vl(); + header.max_size = ZA_PT_SIZE(vq); + + /* If ZA is not active there is only the header */ + if (thread_za_enabled(&target->thread)) + header.size = ZA_PT_SIZE(vq); + else + header.size = ZA_PT_ZA_OFFSET; + + membuf_write(&to, &header, sizeof(header)); + + BUILD_BUG_ON(ZA_PT_ZA_OFFSET != sizeof(header)); + end = ZA_PT_ZA_OFFSET; + + if (target == current) + fpsimd_preserve_current_state(); + + /* Any register data to include? */ + if (thread_za_enabled(&target->thread)) { + start = end; + end = ZA_PT_SIZE(vq); + membuf_write(&to, target->thread.za_state, end - start); + } + + /* Zero any trailing padding */ + start = end; + end = ALIGN(header.size, SVE_VQ_BYTES); + return membuf_zero(&to, end - start); +} + +static int za_set(struct task_struct *target, + const struct user_regset *regset, + unsigned int pos, unsigned int count, + const void *kbuf, const void __user *ubuf) +{ + int ret; + struct user_za_header header; + unsigned int vq; + unsigned long start, end; + + if (!system_supports_sme()) + return -EINVAL; + + /* Header */ + if (count < sizeof(header)) + return -EINVAL; + ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, &header, + 0, sizeof(header)); + if (ret) + goto out; + + /* + * All current ZA_PT_* flags are consumed by + * vec_set_vector_length(), which will also validate them for + * us: + */ + ret = vec_set_vector_length(target, ARM64_VEC_SME, header.vl, + ((unsigned long)header.flags) << 16); + if (ret) + goto out; + + /* Actual VL set may be less than the user asked for: */ + vq = sve_vq_from_vl(task_get_sme_vl(target)); + + /* Ensure there is some SVE storage for streaming mode */ + if (!target->thread.sve_state) { + sve_alloc(target); + if (!target->thread.sve_state) { + clear_thread_flag(TIF_SME); + ret = -ENOMEM; + goto out; + } + } + + /* Allocate/reinit ZA storage */ + sme_alloc(target); + if (!target->thread.za_state) { + ret = -ENOMEM; + clear_tsk_thread_flag(target, TIF_SME); + goto out; + } + + /* If there is no data then disable ZA */ + if (!count) { + target->thread.svcr &= ~SYS_SVCR_EL0_ZA_MASK; + goto out; + } + + /* + * If setting a different VL from the requested VL and there is + * register data, the data layout will be wrong: don't even + * try to set the registers in this case. + */ + if (vq != sve_vq_from_vl(header.vl)) { + ret = -EIO; + goto out; + } + + BUILD_BUG_ON(ZA_PT_ZA_OFFSET != sizeof(header)); + start = ZA_PT_ZA_OFFSET; + end = ZA_PT_SIZE(vq); + ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, + target->thread.za_state, + start, end); + if (ret) + goto out; + + /* Mark ZA as active and let userspace use it */ + set_tsk_thread_flag(target, TIF_SME); + target->thread.svcr |= SYS_SVCR_EL0_ZA_MASK; + +out: + fpsimd_flush_task_state(target); + return ret; +} + #endif /* CONFIG_ARM64_SME */

#ifdef CONFIG_ARM64_PTR_AUTH @@ -1218,6 +1353,7 @@ enum aarch64_regset { #endif #ifdef CONFIG_ARM64_SVE REGSET_SSVE, + REGSET_ZA, #endif #ifdef CONFIG_ARM64_PTR_AUTH REGSET_PAC_MASK, @@ -1309,6 +1445,14 @@ static const struct user_regset aarch64_regsets[] = { .regset_get = ssve_get, .set = ssve_set, }, + [REGSET_ZA] = { /* SME ZA */ + .core_note_type = NT_ARM_ZA, + .n = DIV_ROUND_UP(ZA_PT_ZA_SIZE(SVE_VQ_MAX), SVE_VQ_BYTES), + .size = SVE_VQ_BYTES, + .align = SVE_VQ_BYTES, + .regset_get = za_get, + .set = za_set, + }, #endif #ifdef CONFIG_ARM64_PTR_AUTH [REGSET_PAC_MASK] = { diff --git a/include/uapi/linux/elf.h b/include/uapi/linux/elf.h index a8dc688e1826..97808f958903 100644 --- a/include/uapi/linux/elf.h +++ b/include/uapi/linux/elf.h @@ -432,6 +432,7 @@ typedef struct elf64_shdr { #define NT_ARM_TAGGED_ADDR_CTRL 0x409 /* arm64 tagged address control (prctl()) */ #define NT_ARM_PAC_ENABLED_KEYS 0x40a /* arm64 ptr auth enabled keys (prctl()) */ #define NT_ARM_SSVE 0x40b /* ARM Streaming SVE registers */ +#define NT_ARM_ZA 0x40c /* ARM SME ZA registers */ #define NT_ARC_V2 0x600 /* ARCv2 accumulator/extra registers */ #define NT_VMCOREDD 0x700 /* Vmcore Device Dump Note */ #define NT_MIPS_DSP 0x800 /* MIPS DSP ASE registers */

When saving and restoring the floating point state over an EFI runtime call ensure that we handle streaming mode, only handling FFR if we are not in streaming mode and ensuring that we are in normal mode over the call into runtime services.

We currently assume that ZA will not be modified by runtime services, the specification is not yet finalised so this may need updating if that changes.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/kernel/fpsimd.c | 48 +++++++++++++++++++++++++++++++++----- 1 file changed, 42 insertions(+), 6 deletions(-)

diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 9592cdd7d635..64431bc62472 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -1056,21 +1056,25 @@ int vec_verify_vq_map(enum vec_type type)

static void __init sve_efi_setup(void) { - struct vl_info *info = &vl_info[ARM64_VEC_SVE]; + int max_vl = 0; + int i;

if (!IS_ENABLED(CONFIG_EFI)) return;

+ for (i = 0; i < ARRAY_SIZE(vl_info); i++) + max_vl = max(vl_info[i].max_vl, max_vl); + /* * alloc_percpu() warns and prints a backtrace if this goes wrong. * This is evidence of a crippled system and we are returning void, * so no attempt is made to handle this situation here. */ - if (!sve_vl_valid(info->max_vl)) + if (!sve_vl_valid(max_vl)) goto fail;

efi_sve_state = __alloc_percpu( - SVE_SIG_REGS_SIZE(sve_vq_from_vl(info->max_vl)), SVE_VQ_BYTES); + SVE_SIG_REGS_SIZE(sve_vq_from_vl(max_vl)), SVE_VQ_BYTES); if (!efi_sve_state) goto fail;

@@ -1845,6 +1849,7 @@ EXPORT_SYMBOL(kernel_neon_end); static DEFINE_PER_CPU(struct user_fpsimd_state, efi_fpsimd_state); static DEFINE_PER_CPU(bool, efi_fpsimd_state_used); static DEFINE_PER_CPU(bool, efi_sve_state_used); +static DEFINE_PER_CPU(bool, efi_sm_state);

/* * EFI runtime services support functions @@ -1879,12 +1884,28 @@ void __efi_fpsimd_begin(void) */ if (system_supports_sve() && likely(efi_sve_state)) { char *sve_state = this_cpu_ptr(efi_sve_state); + bool ffr = true; + u64 svcr;

__this_cpu_write(efi_sve_state_used, true);

+ if (system_supports_sme()) { + svcr = read_sysreg_s(SYS_SVCR_EL0); + + if (!system_supports_fa64()) + ffr = svcr & SYS_SVCR_EL0_SM_MASK; + + __this_cpu_write(efi_sm_state, ffr); + } + sve_save_state(sve_state + sve_ffr_offset(sve_max_vl()), &this_cpu_ptr(&efi_fpsimd_state)->fpsr, - true); + ffr); + + if (system_supports_sme()) + sysreg_clear_set_s(SYS_SVCR_EL0, + SYS_SVCR_EL0_SM_MASK, 0); + } else { fpsimd_save_state(this_cpu_ptr(&efi_fpsimd_state)); } @@ -1907,11 +1928,26 @@ void __efi_fpsimd_end(void) if (system_supports_sve() && likely(__this_cpu_read(efi_sve_state_used))) { char const *sve_state = this_cpu_ptr(efi_sve_state); + bool ffr = true; + + /* + * Restore streaming mode; EFI calls are + * normal function calls so should not return in + * streaming mode. + */ + if (system_supports_sme()) { + if (__this_cpu_read(efi_sm_state)) { + sysreg_clear_set_s(SYS_SVCR_EL0, + 0, + SYS_SVCR_EL0_SM_MASK); + if (!system_supports_fa64()) + ffr = efi_sm_state; + } + }

- sve_set_vq(sve_vq_from_vl(sve_get_vl()) - 1); sve_load_state(sve_state + sve_ffr_offset(sve_max_vl()), &this_cpu_ptr(&efi_fpsimd_state)->fpsr, - true); + ffr);

__this_cpu_write(efi_sve_state_used, false); } else {

SME defines two new traps which need to be enabled for guests to ensure that they can't use SME, one for the main SME operations which mirrors the traps for SVE and another for access to TPIDR2 in SCTLR_EL2.

For VHE manage SMEN along with ZEN in activate_traps() and the FP state management callbacks, along with SCTLR_EL2.EnTPIDR2. There is no existing dynamic management of SCTLR_EL2.

For nVHE manage TSM in activate_traps() along with the fine grained traps for TPIDR2 and SMPRI. There is no existing dynamic management of fine grained traps.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/kvm/hyp/nvhe/switch.c | 30 ++++++++++++++++++++++++++++++ arch/arm64/kvm/hyp/vhe/switch.c | 11 ++++++++++- 2 files changed, 40 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/kvm/hyp/nvhe/switch.c b/arch/arm64/kvm/hyp/nvhe/switch.c index 6410d21d8695..caace61ea459 100644 --- a/arch/arm64/kvm/hyp/nvhe/switch.c +++ b/arch/arm64/kvm/hyp/nvhe/switch.c @@ -47,10 +47,24 @@ static void __activate_traps(struct kvm_vcpu *vcpu) val |= CPTR_EL2_TFP | CPTR_EL2_TZ; __activate_traps_fpsimd32(vcpu); } + if (cpus_have_final_cap(ARM64_SME)) + val |= CPTR_EL2_TSM;

write_sysreg(val, cptr_el2); write_sysreg(__this_cpu_read(kvm_hyp_vector), vbar_el2);

+ if (cpus_have_final_cap(ARM64_SME)) { + val = read_sysreg_s(SYS_HFGRTR_EL2); + val &= ~(HFGxTR_EL2_nTPIDR2_EL0_MASK | + HFGxTR_EL2_nSMPRI_EL1_MASK); + write_sysreg_s(val, SYS_HFGRTR_EL2); + + val = read_sysreg_s(SYS_HFGWTR_EL2); + val &= ~(HFGxTR_EL2_nTPIDR2_EL0_MASK | + HFGxTR_EL2_nSMPRI_EL1_MASK); + write_sysreg_s(val, SYS_HFGWTR_EL2); + } + if (cpus_have_final_cap(ARM64_WORKAROUND_SPECULATIVE_AT)) { struct kvm_cpu_context *ctxt = &vcpu->arch.ctxt;

@@ -94,9 +108,25 @@ static void __deactivate_traps(struct kvm_vcpu *vcpu)

write_sysreg(this_cpu_ptr(&kvm_init_params)->hcr_el2, hcr_el2);

+ if (cpus_have_final_cap(ARM64_SME)) { + u64 val; + + val = read_sysreg_s(SYS_HFGRTR_EL2); + val |= HFGxTR_EL2_nTPIDR2_EL0_MASK | + HFGxTR_EL2_nSMPRI_EL1_MASK; + write_sysreg_s(val, SYS_HFGRTR_EL2); + + val = read_sysreg_s(SYS_HFGWTR_EL2); + val |= HFGxTR_EL2_nTPIDR2_EL0_MASK | + HFGxTR_EL2_nSMPRI_EL1_MASK; + write_sysreg_s(val, SYS_HFGWTR_EL2); + } + cptr = CPTR_EL2_DEFAULT; if (vcpu_has_sve(vcpu) && (vcpu->arch.flags & KVM_ARM64_FP_ENABLED)) cptr |= CPTR_EL2_TZ; + if (cpus_have_final_cap(ARM64_SME)) + cptr &= ~CPTR_EL2_TSM;

write_sysreg(cptr, cptr_el2); write_sysreg(__kvm_hyp_host_vector, vbar_el2); diff --git a/arch/arm64/kvm/hyp/vhe/switch.c b/arch/arm64/kvm/hyp/vhe/switch.c index 262dfe03134d..969f20daf97a 100644 --- a/arch/arm64/kvm/hyp/vhe/switch.c +++ b/arch/arm64/kvm/hyp/vhe/switch.c @@ -41,7 +41,8 @@ static void __activate_traps(struct kvm_vcpu *vcpu)

val = read_sysreg(cpacr_el1); val |= CPACR_EL1_TTA; - val &= ~(CPACR_EL1_ZEN_EL0EN | CPACR_EL1_ZEN_EL1EN); + val &= ~(CPACR_EL1_ZEN_EL0EN | CPACR_EL1_ZEN_EL1EN | + CPACR_EL1_SMEN_EL0EN | CPACR_EL1_SMEN_EL1EN);

/* * With VHE (HCR.E2H == 1), accesses to CPACR_EL1 are routed to @@ -62,6 +63,10 @@ static void __activate_traps(struct kvm_vcpu *vcpu) __activate_traps_fpsimd32(vcpu); }

+ if (cpus_have_final_cap(ARM64_SME)) + write_sysreg(read_sysreg(sctlr_el2) & ~SCTLR_ELx_ENTP2, + sctlr_el2); + write_sysreg(val, cpacr_el1);

write_sysreg(__this_cpu_read(kvm_hyp_vector), vbar_el1); @@ -83,6 +88,10 @@ static void __deactivate_traps(struct kvm_vcpu *vcpu) */ asm(ALTERNATIVE("nop", "isb", ARM64_WORKAROUND_SPECULATIVE_AT));

+ if (cpus_have_final_cap(ARM64_SME)) + write_sysreg(read_sysreg(sctlr_el2) | SCTLR_ELx_ENTP2, + sctlr_el2); + write_sysreg(CPACR_EL1_DEFAULT, cpacr_el1);

if (!arm64_kernel_unmapped_at_el0())

Now that basline support for the Scalable Matrix Extension (SME) is present introduce the Kconfig option allowing it to be built. While the feature registers don't impose a strong requirement for a system with SME to support SVE at runtime the support for streaming mode SVE is mostly shared with normal SVE so depend on SVE.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/Kconfig | 11 +++++++++++ 1 file changed, 11 insertions(+)

diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig index 57c4c995965f..0897984918e8 100644 --- a/arch/arm64/Kconfig +++ b/arch/arm64/Kconfig @@ -1939,6 +1939,17 @@ config ARM64_SVE booting the kernel. If unsure and you are not observing these symptoms, you should assume that it is safe to say Y.

+config ARM64_SME + bool "ARM Scalable Matrix Extension support" + default y + depends on ARM64_SVE + help + The Scalable Matrix Extension (SME) is an extension to the AArch64 + execution state which utilises a substantial subset of the SVE + instruction set, together with the addition of new architectural + register state capable of holding two dimensional matrix tiles to + enable various matrix operations. + config ARM64_MODULE_PLTS bool "Use PLTs to allow module memory to spill over into vmalloc area" depends on MODULES

The Scalable Matrix Extenions (SME) introduces additional register state with configurable vector lengths, similar to SVE but configured separately. Extend vlset to support configuring this state with a --sme or -s command line option.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Shuah Khan skhan@linuxfoundation.org Acked-by: Catalin Marinas catalin.marinas@arm.com --- tools/testing/selftests/arm64/fp/vlset.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/tools/testing/selftests/arm64/fp/vlset.c b/tools/testing/selftests/arm64/fp/vlset.c index 308d27a68226..76912a581a95 100644 --- a/tools/testing/selftests/arm64/fp/vlset.c +++ b/tools/testing/selftests/arm64/fp/vlset.c @@ -22,12 +22,15 @@ static int inherit = 0; static int no_inherit = 0; static int force = 0; static unsigned long vl; +static int set_ctl = PR_SVE_SET_VL; +static int get_ctl = PR_SVE_GET_VL;

static const struct option options[] = { { "force", no_argument, NULL, 'f' }, { "inherit", no_argument, NULL, 'i' }, { "max", no_argument, NULL, 'M' }, { "no-inherit", no_argument, &no_inherit, 1 }, + { "sme", no_argument, NULL, 's' }, { "help", no_argument, NULL, '?' }, {} }; @@ -50,6 +53,9 @@ static int parse_options(int argc, char **argv) case 'M': vl = SVE_VL_MAX; break; case 'f': force = 1; break; case 'i': inherit = 1; break; + case 's': set_ctl = PR_SME_SET_VL; + get_ctl = PR_SME_GET_VL; + break; case 0: break; default: goto error; } @@ -125,14 +131,14 @@ int main(int argc, char **argv) if (inherit) flags |= PR_SVE_VL_INHERIT;

- t = prctl(PR_SVE_SET_VL, vl | flags); + t = prctl(set_ctl, vl | flags); if (t < 0) { fprintf(stderr, "%s: PR_SVE_SET_VL: %s\n", program_name, strerror(errno)); goto error; }

- t = prctl(PR_SVE_GET_VL); + t = prctl(get_ctl); if (t == -1) { fprintf(stderr, "%s: PR_SVE_GET_VL: %s\n", program_name, strerror(errno));

Provide RDVL helpers for SME and extend the main vector configuration tests to cover SME.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Shuah Khan skhan@linuxfoundation.org Acked-by: Catalin Marinas catalin.marinas@arm.com --- tools/testing/selftests/arm64/fp/.gitignore | 1 + tools/testing/selftests/arm64/fp/Makefile | 3 ++- tools/testing/selftests/arm64/fp/rdvl-sme.c | 14 ++++++++++++++ tools/testing/selftests/arm64/fp/rdvl.S | 10 ++++++++++ tools/testing/selftests/arm64/fp/rdvl.h | 1 + tools/testing/selftests/arm64/fp/vec-syscfg.c | 10 ++++++++++ 6 files changed, 38 insertions(+), 1 deletion(-) create mode 100644 tools/testing/selftests/arm64/fp/rdvl-sme.c

diff --git a/tools/testing/selftests/arm64/fp/.gitignore b/tools/testing/selftests/arm64/fp/.gitignore index c50d86331ed2..6e9a610c5e5d 100644 --- a/tools/testing/selftests/arm64/fp/.gitignore +++ b/tools/testing/selftests/arm64/fp/.gitignore @@ -1,5 +1,6 @@ fp-pidbench fpsimd-test +rdvl-sme rdvl-sve sve-probe-vls sve-ptrace diff --git a/tools/testing/selftests/arm64/fp/Makefile b/tools/testing/selftests/arm64/fp/Makefile index 95f0b877a060..a224fff8082b 100644 --- a/tools/testing/selftests/arm64/fp/Makefile +++ b/tools/testing/selftests/arm64/fp/Makefile @@ -3,7 +3,7 @@ CFLAGS += -I../../../../../usr/include/ TEST_GEN_PROGS := sve-ptrace sve-probe-vls vec-syscfg TEST_PROGS_EXTENDED := fp-pidbench fpsimd-test fpsimd-stress \ - rdvl-sve \ + rdvl-sme rdvl-sve \ sve-test sve-stress \ vlset

@@ -13,6 +13,7 @@ fp-pidbench: fp-pidbench.S asm-utils.o $(CC) -nostdlib $^ -o $@ fpsimd-test: fpsimd-test.o asm-utils.o $(CC) -nostdlib $^ -o $@ +rdvl-sme: rdvl-sme.o rdvl.o rdvl-sve: rdvl-sve.o rdvl.o sve-ptrace: sve-ptrace.o sve-probe-vls: sve-probe-vls.o rdvl.o diff --git a/tools/testing/selftests/arm64/fp/rdvl-sme.c b/tools/testing/selftests/arm64/fp/rdvl-sme.c new file mode 100644 index 000000000000..49b0b2e08bac --- /dev/null +++ b/tools/testing/selftests/arm64/fp/rdvl-sme.c @@ -0,0 +1,14 @@ +// SPDX-License-Identifier: GPL-2.0-only + +#include <stdio.h> + +#include "rdvl.h" + +int main(void) +{ + int vl = rdvl_sme(); + + printf("%d\n", vl); + + return 0; +} diff --git a/tools/testing/selftests/arm64/fp/rdvl.S b/tools/testing/selftests/arm64/fp/rdvl.S index c916c1c9defd..20dc29996dc6 100644 --- a/tools/testing/selftests/arm64/fp/rdvl.S +++ b/tools/testing/selftests/arm64/fp/rdvl.S @@ -1,6 +1,8 @@ // SPDX-License-Identifier: GPL-2.0-only // Copyright (C) 2021 ARM Limited.

+#include "sme-inst.h" + .arch_extension sve

.globl rdvl_sve @@ -8,3 +10,11 @@ rdvl_sve: hint 34 // BTI C rdvl x0, #1 ret + +.globl rdvl_sme +rdvl_sme: + hint 34 // BTI C + + rdsvl 0, 1 + + ret diff --git a/tools/testing/selftests/arm64/fp/rdvl.h b/tools/testing/selftests/arm64/fp/rdvl.h index 7c9d953fc9e7..5d323679fbc9 100644 --- a/tools/testing/selftests/arm64/fp/rdvl.h +++ b/tools/testing/selftests/arm64/fp/rdvl.h @@ -3,6 +3,7 @@ #ifndef RDVL_H #define RDVL_H

+int rdvl_sme(void); int rdvl_sve(void);

#endif diff --git a/tools/testing/selftests/arm64/fp/vec-syscfg.c b/tools/testing/selftests/arm64/fp/vec-syscfg.c index c90658811a83..9bcfcdc34ee9 100644 --- a/tools/testing/selftests/arm64/fp/vec-syscfg.c +++ b/tools/testing/selftests/arm64/fp/vec-syscfg.c @@ -51,6 +51,16 @@ static struct vec_data vec_data[] = { .prctl_set = PR_SVE_SET_VL, .default_vl_file = "/proc/sys/abi/sve_default_vector_length", }, + { + .name = "SME", + .hwcap_type = AT_HWCAP2, + .hwcap = HWCAP2_SME, + .rdvl = rdvl_sme, + .rdvl_binary = "./rdvl-sme", + .prctl_get = PR_SME_GET_VL, + .prctl_set = PR_SME_SET_VL, + .default_vl_file = "/proc/sys/abi/sme_default_vector_length", + }, };

static int stdio_read_integer(FILE *f, const char *what, int *val)

As part of the generic code for signal handling test cases we parse all signal frames to make sure they have at least the basic form we expect and that there are no unexpected frames present in the signal context. Add coverage of the ZA signal frame to this code.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Shuah Khan skhan@linuxfoundation.org Acked-by: Catalin Marinas catalin.marinas@arm.com --- .../arm64/signal/testcases/testcases.c | 36 +++++++++++++++++++ .../arm64/signal/testcases/testcases.h | 3 +- 2 files changed, 38 insertions(+), 1 deletion(-)

diff --git a/tools/testing/selftests/arm64/signal/testcases/testcases.c b/tools/testing/selftests/arm64/signal/testcases/testcases.c index 8c2a57fc2f9c..84c36bee4d82 100644 --- a/tools/testing/selftests/arm64/signal/testcases/testcases.c +++ b/tools/testing/selftests/arm64/signal/testcases/testcases.c @@ -75,6 +75,31 @@ bool validate_sve_context(struct sve_context *sve, char **err) return true; }

+bool validate_za_context(struct za_context *za, char **err) +{ + /* Size will be rounded up to a multiple of 16 bytes */ + size_t regs_size + = ((ZA_SIG_CONTEXT_SIZE(sve_vq_from_vl(za->vl)) + 15) / 16) * 16; + + if (!za || !err) + return false; + + /* Either a bare za_context or a za_context followed by regs data */ + if ((za->head.size != sizeof(struct za_context)) && + (za->head.size != regs_size)) { + *err = "bad size for ZA context"; + return false; + } + + if (!sve_vl_valid(za->vl)) { + *err = "SME VL in ZA context invalid"; + + return false; + } + + return true; +} + bool validate_reserved(ucontext_t *uc, size_t resv_sz, char **err) { bool terminated = false; @@ -82,6 +107,7 @@ bool validate_reserved(ucontext_t *uc, size_t resv_sz, char **err) int flags = 0; struct extra_context *extra = NULL; struct sve_context *sve = NULL; + struct za_context *za = NULL; struct _aarch64_ctx *head = (struct _aarch64_ctx *)uc->uc_mcontext.__reserved;

@@ -120,6 +146,13 @@ bool validate_reserved(ucontext_t *uc, size_t resv_sz, char **err) sve = (struct sve_context *)head; flags |= SVE_CTX; break; + case ZA_MAGIC: + if (flags & ZA_CTX) + *err = "Multiple ZA_MAGIC"; + /* Size is validated in validate_za_context() */ + za = (struct za_context *)head; + flags |= ZA_CTX; + break; case EXTRA_MAGIC: if (flags & EXTRA_CTX) *err = "Multiple EXTRA_MAGIC"; @@ -165,6 +198,9 @@ bool validate_reserved(ucontext_t *uc, size_t resv_sz, char **err) if (flags & SVE_CTX) if (!validate_sve_context(sve, err)) return false; + if (flags & ZA_CTX) + if (!validate_za_context(za, err)) + return false;

head = GET_RESV_NEXT_HEAD(head); } diff --git a/tools/testing/selftests/arm64/signal/testcases/testcases.h b/tools/testing/selftests/arm64/signal/testcases/testcases.h index ad884c135314..49f1d5de7b5b 100644 --- a/tools/testing/selftests/arm64/signal/testcases/testcases.h +++ b/tools/testing/selftests/arm64/signal/testcases/testcases.h @@ -16,7 +16,8 @@

#define FPSIMD_CTX (1 << 0) #define SVE_CTX (1 << 1) -#define EXTRA_CTX (1 << 2) +#define ZA_CTX (1 << 2) +#define EXTRA_CTX (1 << 3)

#define KSFT_BAD_MAGIC 0xdeadbeef

Add test cases for the SME signal handing ABI patterned off the SVE tests. Due to the small size of the tests and the differences in ABI (especially around needing to account for both streaming SVE and ZA) there is some code duplication here.

We currently cover: - Reporting of the vector length. - Lack of support for changing vector length. - Presence and size of register state for streaming SVE and ZA.

As with the SVE tests we do not yet have any validation of register contents.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Shuah Khan skhan@linuxfoundation.org Acked-by: Catalin Marinas catalin.marinas@arm.com --- .../testing/selftests/arm64/signal/.gitignore | 3 + .../selftests/arm64/signal/test_signals.h | 4 + .../arm64/signal/test_signals_utils.c | 6 + .../testcases/fake_sigreturn_sme_change_vl.c | 92 ++++++++++++ .../arm64/signal/testcases/sme_trap_no_sm.c | 38 +++++ .../signal/testcases/sme_trap_non_streaming.c | 45 ++++++ .../arm64/signal/testcases/sme_trap_za.c | 36 +++++ .../selftests/arm64/signal/testcases/sme_vl.c | 68 +++++++++ .../arm64/signal/testcases/ssve_regs.c | 135 ++++++++++++++++++ .../arm64/signal/testcases/za_regs.c | 128 +++++++++++++++++ 10 files changed, 555 insertions(+) create mode 100644 tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_sme_change_vl.c create mode 100644 tools/testing/selftests/arm64/signal/testcases/sme_trap_no_sm.c create mode 100644 tools/testing/selftests/arm64/signal/testcases/sme_trap_non_streaming.c create mode 100644 tools/testing/selftests/arm64/signal/testcases/sme_trap_za.c create mode 100644 tools/testing/selftests/arm64/signal/testcases/sme_vl.c create mode 100644 tools/testing/selftests/arm64/signal/testcases/ssve_regs.c create mode 100644 tools/testing/selftests/arm64/signal/testcases/za_regs.c

diff --git a/tools/testing/selftests/arm64/signal/.gitignore b/tools/testing/selftests/arm64/signal/.gitignore index c1742755abb9..e8d2b57f73ec 100644 --- a/tools/testing/selftests/arm64/signal/.gitignore +++ b/tools/testing/selftests/arm64/signal/.gitignore @@ -1,5 +1,8 @@ # SPDX-License-Identifier: GPL-2.0-only mangle_* fake_sigreturn_* +sme_* +ssve_* sve_* +za_* !*.[ch] diff --git a/tools/testing/selftests/arm64/signal/test_signals.h b/tools/testing/selftests/arm64/signal/test_signals.h index f909b70d9e98..c70fdec7d7c4 100644 --- a/tools/testing/selftests/arm64/signal/test_signals.h +++ b/tools/testing/selftests/arm64/signal/test_signals.h @@ -34,11 +34,15 @@ enum { FSSBS_BIT, FSVE_BIT, + FSME_BIT, + FSME_FA64_BIT, FMAX_END };

#define FEAT_SSBS (1UL << FSSBS_BIT) #define FEAT_SVE (1UL << FSVE_BIT) +#define FEAT_SME (1UL << FSME_BIT) +#define FEAT_SME_FA64 (1UL << FSME_FA64_BIT)

/* * A descriptor used to describe and configure a test case. diff --git a/tools/testing/selftests/arm64/signal/test_signals_utils.c b/tools/testing/selftests/arm64/signal/test_signals_utils.c index 5743897984b0..b588d10afd5b 100644 --- a/tools/testing/selftests/arm64/signal/test_signals_utils.c +++ b/tools/testing/selftests/arm64/signal/test_signals_utils.c @@ -27,6 +27,8 @@ static int sig_copyctx = SIGTRAP; static char const *const feats_names[FMAX_END] = { " SSBS ", " SVE ", + " SME ", + " FA64 ", };

#define MAX_FEATS_SZ 128 @@ -268,6 +270,10 @@ int test_init(struct tdescr *td) td->feats_supported |= FEAT_SSBS; if (getauxval(AT_HWCAP) & HWCAP_SVE) td->feats_supported |= FEAT_SVE; + if (getauxval(AT_HWCAP2) & HWCAP2_SME) + td->feats_supported |= FEAT_SME; + if (getauxval(AT_HWCAP2) & HWCAP2_SME_FA64) + td->feats_supported |= FEAT_SME_FA64; if (feats_ok(td)) { if (td->feats_required & td->feats_supported) fprintf(stderr, diff --git a/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_sme_change_vl.c b/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_sme_change_vl.c new file mode 100644 index 000000000000..7ed762b7202f --- /dev/null +++ b/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_sme_change_vl.c @@ -0,0 +1,92 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) 2021 ARM Limited + * + * Attempt to change the streaming SVE vector length in a signal + * handler, this is not supported and is expected to segfault. + */ + +#include <signal.h> +#include <ucontext.h> +#include <sys/prctl.h> + +#include "test_signals_utils.h" +#include "testcases.h" + +struct fake_sigframe sf; +static unsigned int vls[SVE_VQ_MAX]; +unsigned int nvls = 0; + +static bool sme_get_vls(struct tdescr *td) +{ + int vq, vl; + + /* + * Enumerate up to SVE_VQ_MAX vector lengths + */ + for (vq = SVE_VQ_MAX; vq > 0; --vq) { + vl = prctl(PR_SVE_SET_VL, vq * 16); + if (vl == -1) + return false; + + vl &= PR_SME_VL_LEN_MASK; + + /* Skip missing VLs */ + vq = sve_vq_from_vl(vl); + + vls[nvls++] = vl; + } + + /* We need at least two VLs */ + if (nvls < 2) { + fprintf(stderr, "Only %d VL supported\n", nvls); + return false; + } + + return true; +} + +static int fake_sigreturn_ssve_change_vl(struct tdescr *td, + siginfo_t *si, ucontext_t *uc) +{ + size_t resv_sz, offset; + struct _aarch64_ctx *head = GET_SF_RESV_HEAD(sf); + struct sve_context *sve; + + /* Get a signal context with a SME ZA frame in it */ + if (!get_current_context(td, &sf.uc)) + return 1; + + resv_sz = GET_SF_RESV_SIZE(sf); + head = get_header(head, SVE_MAGIC, resv_sz, &offset); + if (!head) { + fprintf(stderr, "No SVE context\n"); + return 1; + } + + if (head->size != sizeof(struct sve_context)) { + fprintf(stderr, "Register data present, aborting\n"); + return 1; + } + + sve = (struct sve_context *)head; + + /* No changes are supported; init left us at minimum VL so go to max */ + fprintf(stderr, "Attempting to change VL from %d to %d\n", + sve->vl, vls[0]); + sve->vl = vls[0]; + + fake_sigreturn(&sf, sizeof(sf), 0); + + return 1; +} + +struct tdescr tde = { + .name = "FAKE_SIGRETURN_SSVE_CHANGE", + .descr = "Attempt to change Streaming SVE VL", + .feats_required = FEAT_SME, + .sig_ok = SIGSEGV, + .timeout = 3, + .init = sme_get_vls, + .run = fake_sigreturn_ssve_change_vl, +}; diff --git a/tools/testing/selftests/arm64/signal/testcases/sme_trap_no_sm.c b/tools/testing/selftests/arm64/signal/testcases/sme_trap_no_sm.c new file mode 100644 index 000000000000..f9d76ae32bba --- /dev/null +++ b/tools/testing/selftests/arm64/signal/testcases/sme_trap_no_sm.c @@ -0,0 +1,38 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) 2021 ARM Limited + * + * Verify that using a streaming mode instruction without enabling it + * generates a SIGILL. + */ + +#include <signal.h> +#include <ucontext.h> +#include <sys/prctl.h> + +#include "test_signals_utils.h" +#include "testcases.h" + +int sme_trap_no_sm_trigger(struct tdescr *td) +{ + /* SMSTART ZA ; ADDHA ZA0.S, P0/M, P0/M, Z0.S */ + asm volatile(".inst 0xd503457f ; .inst 0xc0900000"); + + return 0; +} + +int sme_trap_no_sm_run(struct tdescr *td, siginfo_t *si, ucontext_t *uc) +{ + return 1; +} + +struct tdescr tde = { + .name = "SME trap without SM", + .descr = "Check that we get a SIGILL if we use streaming mode without enabling it", + .timeout = 3, + .feats_required = FEAT_SME, /* We need a SMSTART ZA */ + .sanity_disabled = true, + .trigger = sme_trap_no_sm_trigger, + .run = sme_trap_no_sm_run, + .sig_ok = SIGILL, +}; diff --git a/tools/testing/selftests/arm64/signal/testcases/sme_trap_non_streaming.c b/tools/testing/selftests/arm64/signal/testcases/sme_trap_non_streaming.c new file mode 100644 index 000000000000..e469ae5348e3 --- /dev/null +++ b/tools/testing/selftests/arm64/signal/testcases/sme_trap_non_streaming.c @@ -0,0 +1,45 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) 2021 ARM Limited + * + * Verify that using an instruction not supported in streaming mode + * traps when in streaming mode. + */ + +#include <signal.h> +#include <ucontext.h> +#include <sys/prctl.h> + +#include "test_signals_utils.h" +#include "testcases.h" + +int sme_trap_non_streaming_trigger(struct tdescr *td) +{ + /* + * The framework will handle SIGILL so we need to exit SM to + * stop any other code triggering a further SIGILL down the + * line from using a streaming-illegal instruction. + */ + asm volatile(".inst 0xd503437f; /* SMSTART ZA */ \ + cnt v0.16b, v0.16b; \ + .inst 0xd503447f /* SMSTOP ZA */"); + + return 0; +} + +int sme_trap_non_streaming_run(struct tdescr *td, siginfo_t *si, ucontext_t *uc) +{ + return 1; +} + +struct tdescr tde = { + .name = "SME SM trap unsupported instruction", + .descr = "Check that we get a SIGILL if we use an unsupported instruction in streaming mode", + .feats_required = FEAT_SME, + .feats_incompatible = FEAT_SME_FA64, + .timeout = 3, + .sanity_disabled = true, + .trigger = sme_trap_non_streaming_trigger, + .run = sme_trap_non_streaming_run, + .sig_ok = SIGILL, +}; diff --git a/tools/testing/selftests/arm64/signal/testcases/sme_trap_za.c b/tools/testing/selftests/arm64/signal/testcases/sme_trap_za.c new file mode 100644 index 000000000000..3a7747af4715 --- /dev/null +++ b/tools/testing/selftests/arm64/signal/testcases/sme_trap_za.c @@ -0,0 +1,36 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) 2021 ARM Limited + * + * Verify that accessing ZA without enabling it generates a SIGILL. + */ + +#include <signal.h> +#include <ucontext.h> +#include <sys/prctl.h> + +#include "test_signals_utils.h" +#include "testcases.h" + +int sme_trap_za_trigger(struct tdescr *td) +{ + /* ZERO ZA */ + asm volatile(".inst 0xc00800ff"); + + return 0; +} + +int sme_trap_za_run(struct tdescr *td, siginfo_t *si, ucontext_t *uc) +{ + return 1; +} + +struct tdescr tde = { + .name = "SME ZA trap", + .descr = "Check that we get a SIGILL if we access ZA without enabling", + .timeout = 3, + .sanity_disabled = true, + .trigger = sme_trap_za_trigger, + .run = sme_trap_za_run, + .sig_ok = SIGILL, +}; diff --git a/tools/testing/selftests/arm64/signal/testcases/sme_vl.c b/tools/testing/selftests/arm64/signal/testcases/sme_vl.c new file mode 100644 index 000000000000..13ff3b35cbaf --- /dev/null +++ b/tools/testing/selftests/arm64/signal/testcases/sme_vl.c @@ -0,0 +1,68 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) 2021 ARM Limited + * + * Check that the SME vector length reported in signal contexts is the + * expected one. + */ + +#include <signal.h> +#include <ucontext.h> +#include <sys/prctl.h> + +#include "test_signals_utils.h" +#include "testcases.h" + +struct fake_sigframe sf; +unsigned int vl; + +static bool get_sme_vl(struct tdescr *td) +{ + int ret = prctl(PR_SME_GET_VL); + if (ret == -1) + return false; + + vl = ret; + + return true; +} + +static int sme_vl(struct tdescr *td, siginfo_t *si, ucontext_t *uc) +{ + size_t resv_sz, offset; + struct _aarch64_ctx *head = GET_SF_RESV_HEAD(sf); + struct za_context *za; + + /* Get a signal context which should have a ZA frame in it */ + if (!get_current_context(td, &sf.uc)) + return 1; + + resv_sz = GET_SF_RESV_SIZE(sf); + head = get_header(head, ZA_MAGIC, resv_sz, &offset); + if (!head) { + fprintf(stderr, "No ZA context\n"); + return 1; + } + za = (struct za_context *)head; + + if (za->vl != vl) { + fprintf(stderr, "ZA sigframe VL %u, expected %u\n", + za->vl, vl); + return 1; + } else { + fprintf(stderr, "got expected VL %u\n", vl); + } + + td->pass = 1; + + return 0; +} + +struct tdescr tde = { + .name = "SME VL", + .descr = "Check that we get the right SME VL reported", + .feats_required = FEAT_SME, + .timeout = 3, + .init = get_sme_vl, + .run = sme_vl, +}; diff --git a/tools/testing/selftests/arm64/signal/testcases/ssve_regs.c b/tools/testing/selftests/arm64/signal/testcases/ssve_regs.c new file mode 100644 index 000000000000..9022a6cab4b3 --- /dev/null +++ b/tools/testing/selftests/arm64/signal/testcases/ssve_regs.c @@ -0,0 +1,135 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) 2021 ARM Limited + * + * Verify that the streaming SVE register context in signal frames is + * set up as expected. + */ + +#include <signal.h> +#include <ucontext.h> +#include <sys/prctl.h> + +#include "test_signals_utils.h" +#include "testcases.h" + +struct fake_sigframe sf; +static unsigned int vls[SVE_VQ_MAX]; +unsigned int nvls = 0; + +static bool sme_get_vls(struct tdescr *td) +{ + int vq, vl; + + /* + * Enumerate up to SVE_VQ_MAX vector lengths + */ + for (vq = SVE_VQ_MAX; vq > 0; --vq) { + vl = prctl(PR_SME_SET_VL, vq * 16); + if (vl == -1) + return false; + + vl &= PR_SME_VL_LEN_MASK; + + /* Skip missing VLs */ + vq = sve_vq_from_vl(vl); + + vls[nvls++] = vl; + } + + /* We need at least one VL */ + if (nvls < 1) { + fprintf(stderr, "Only %d VL supported\n", nvls); + return false; + } + + return true; +} + +static void setup_ssve_regs(void) +{ + /* smstart sm; real data is TODO */ + asm volatile(".inst 0xd503437f" : : : ); +} + +static int do_one_sme_vl(struct tdescr *td, siginfo_t *si, ucontext_t *uc, + unsigned int vl) +{ + size_t resv_sz, offset; + struct _aarch64_ctx *head = GET_SF_RESV_HEAD(sf); + struct sve_context *ssve; + int ret; + + fprintf(stderr, "Testing VL %d\n", vl); + + ret = prctl(PR_SME_SET_VL, vl); + if (ret != vl) { + fprintf(stderr, "Failed to set VL, got %d\n", ret); + return 1; + } + + /* + * Get a signal context which should have a SVE frame and registers + * in it. + */ + setup_ssve_regs(); + if (!get_current_context(td, &sf.uc)) + return 1; + + resv_sz = GET_SF_RESV_SIZE(sf); + head = get_header(head, SVE_MAGIC, resv_sz, &offset); + if (!head) { + fprintf(stderr, "No SVE context\n"); + return 1; + } + + ssve = (struct sve_context *)head; + if (ssve->vl != vl) { + fprintf(stderr, "Got VL %d, expected %d\n", ssve->vl, vl); + return 1; + } + + /* The actual size validation is done in get_current_context() */ + fprintf(stderr, "Got expected size %u and VL %d\n", + head->size, ssve->vl); + + return 0; +} + +static int sme_regs(struct tdescr *td, siginfo_t *si, ucontext_t *uc) +{ + int i; + + for (i = 0; i < nvls; i++) { + /* + * TODO: the signal test helpers can't currently cope + * with signal frames bigger than struct sigcontext, + * skip VLs that will trigger that. + */ + if (vls[i] > 64) { + printf("Skipping VL %u due to stack size\n", vls[i]); + continue; + } + + if (do_one_sme_vl(td, si, uc, vls[i])) + return 1; + } + + td->pass = 1; + + return 0; +} + +struct tdescr tde = { + .name = "Streaming SVE registers", + .descr = "Check that we get the right Streaming SVE registers reported", + /* + * We shouldn't require FA64 but things like memset() used in the + * helpers might use unsupported instructions so for now disable + * the test unless we've got the full instruction set. + */ + .feats_required = FEAT_SME | FEAT_SME_FA64, + .timeout = 3, + .init = sme_get_vls, + .run = sme_regs, +}; diff --git a/tools/testing/selftests/arm64/signal/testcases/za_regs.c b/tools/testing/selftests/arm64/signal/testcases/za_regs.c new file mode 100644 index 000000000000..b94e4f99fcac --- /dev/null +++ b/tools/testing/selftests/arm64/signal/testcases/za_regs.c @@ -0,0 +1,128 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) 2021 ARM Limited + * + * Verify that the ZA register context in signal frames is set up as + * expected. + */ + +#include <signal.h> +#include <ucontext.h> +#include <sys/prctl.h> + +#include "test_signals_utils.h" +#include "testcases.h" + +struct fake_sigframe sf; +static unsigned int vls[SVE_VQ_MAX]; +unsigned int nvls = 0; + +static bool sme_get_vls(struct tdescr *td) +{ + int vq, vl; + + /* + * Enumerate up to SVE_VQ_MAX vector lengths + */ + for (vq = SVE_VQ_MAX; vq > 0; --vq) { + vl = prctl(PR_SVE_SET_VL, vq * 16); + if (vl == -1) + return false; + + vl &= PR_SME_VL_LEN_MASK; + + /* Skip missing VLs */ + vq = sve_vq_from_vl(vl); + + vls[nvls++] = vl; + } + + /* We need at least one VL */ + if (nvls < 1) { + fprintf(stderr, "Only %d VL supported\n", nvls); + return false; + } + + return true; +} + +static void setup_za_regs(void) +{ + /* smstart za; real data is TODO */ + asm volatile(".inst 0xd503457f" : : : ); +} + +static int do_one_sme_vl(struct tdescr *td, siginfo_t *si, ucontext_t *uc, + unsigned int vl) +{ + size_t resv_sz, offset; + struct _aarch64_ctx *head = GET_SF_RESV_HEAD(sf); + struct za_context *za; + + fprintf(stderr, "Testing VL %d\n", vl); + + if (prctl(PR_SME_SET_VL, vl) != vl) { + fprintf(stderr, "Failed to set VL\n"); + return 1; + } + + /* + * Get a signal context which should have a SVE frame and registers + * in it. + */ + setup_za_regs(); + if (!get_current_context(td, &sf.uc)) + return 1; + + resv_sz = GET_SF_RESV_SIZE(sf); + head = get_header(head, ZA_MAGIC, resv_sz, &offset); + if (!head) { + fprintf(stderr, "No ZA context\n"); + return 1; + } + + za = (struct za_context *)head; + if (za->vl != vl) { + fprintf(stderr, "Got VL %d, expected %d\n", za->vl, vl); + return 1; + } + + /* The actual size validation is done in get_current_context() */ + fprintf(stderr, "Got expected size %u and VL %d\n", + head->size, za->vl); + + return 0; +} + +static int sme_regs(struct tdescr *td, siginfo_t *si, ucontext_t *uc) +{ + int i; + + for (i = 0; i < nvls; i++) { + /* + * TODO: the signal test helpers can't currently cope + * with signal frames bigger than struct sigcontext, + * skip VLs that will trigger that. + */ + if (vls[i] > 32) { + printf("Skipping VL %u due to stack size\n", vls[i]); + continue; + } + + if (do_one_sme_vl(td, si, uc, vls[i])) + return 1; + } + + td->pass = 1; + + return 0; +} + +struct tdescr tde = { + .name = "ZA register", + .descr = "Check that we get the right ZA registers reported", + .feats_required = FEAT_SME, + .timeout = 3, + .init = sme_get_vls, + .run = sme_regs, +};

Add some basic coverage for the ZA ptrace interface, including walking through all the vector lengths supported in the system. Unlike SVE doing syscalls does not discard the ZA state so when we set data in ZA we run the child process briefly, having it add one to each byte in ZA in order to validate that both the vector size and data are being read and written as expected when the process runs.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Shuah Khan skhan@linuxfoundation.org Acked-by: Catalin Marinas catalin.marinas@arm.com --- tools/testing/selftests/arm64/fp/.gitignore | 1 + tools/testing/selftests/arm64/fp/Makefile | 3 +- tools/testing/selftests/arm64/fp/za-ptrace.c | 356 +++++++++++++++++++ 3 files changed, 359 insertions(+), 1 deletion(-) create mode 100644 tools/testing/selftests/arm64/fp/za-ptrace.c

diff --git a/tools/testing/selftests/arm64/fp/.gitignore b/tools/testing/selftests/arm64/fp/.gitignore index ead3197e720b..d98d3d48b504 100644 --- a/tools/testing/selftests/arm64/fp/.gitignore +++ b/tools/testing/selftests/arm64/fp/.gitignore @@ -8,4 +8,5 @@ sve-test ssve-test vec-syscfg vlset +za-ptrace za-test diff --git a/tools/testing/selftests/arm64/fp/Makefile b/tools/testing/selftests/arm64/fp/Makefile index 38d2d0d5a0eb..807a8faf8d57 100644 --- a/tools/testing/selftests/arm64/fp/Makefile +++ b/tools/testing/selftests/arm64/fp/Makefile @@ -1,7 +1,7 @@ # SPDX-License-Identifier: GPL-2.0

CFLAGS += -I../../../../../usr/include/ -TEST_GEN_PROGS := sve-ptrace sve-probe-vls vec-syscfg +TEST_GEN_PROGS := sve-ptrace sve-probe-vls vec-syscfg za-ptrace TEST_PROGS_EXTENDED := fp-pidbench fpsimd-test fpsimd-stress \ rdvl-sme rdvl-sve \ sve-test sve-stress \ @@ -27,5 +27,6 @@ vec-syscfg: vec-syscfg.o rdvl.o vlset: vlset.o za-test: za-test.o asm-utils.o $(CC) -nostdlib $^ -o $@ +za-ptrace: za-ptrace.o

include ../../lib.mk diff --git a/tools/testing/selftests/arm64/fp/za-ptrace.c b/tools/testing/selftests/arm64/fp/za-ptrace.c new file mode 100644 index 000000000000..bf6158654056 --- /dev/null +++ b/tools/testing/selftests/arm64/fp/za-ptrace.c @@ -0,0 +1,356 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Copyright (C) 2021 ARM Limited. + */ +#include <errno.h> +#include <stdbool.h> +#include <stddef.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <unistd.h> +#include <sys/auxv.h> +#include <sys/prctl.h> +#include <sys/ptrace.h> +#include <sys/types.h> +#include <sys/uio.h> +#include <sys/wait.h> +#include <asm/sigcontext.h> +#include <asm/ptrace.h> + +#include "../../kselftest.h" + +/* <linux/elf.h> and <sys/auxv.h> don't like each other, so: */ +#ifndef NT_ARM_ZA +#define NT_ARM_ZA 0x40c +#endif + +#define EXPECTED_TESTS (((SVE_VQ_MAX - SVE_VQ_MIN) + 1) * 3) + +static void fill_buf(char *buf, size_t size) +{ + int i; + + for (i = 0; i < size; i++) + buf[i] = random(); +} + +static int do_child(void) +{ + if (ptrace(PTRACE_TRACEME, -1, NULL, NULL)) + ksft_exit_fail_msg("PTRACE_TRACEME", strerror(errno)); + + if (raise(SIGSTOP)) + ksft_exit_fail_msg("raise(SIGSTOP)", strerror(errno)); + + return EXIT_SUCCESS; +} + +static struct user_za_header *get_za(pid_t pid, void **buf, size_t *size) +{ + struct user_za_header *za; + void *p; + size_t sz = sizeof(*za); + struct iovec iov; + + while (1) { + if (*size < sz) { + p = realloc(*buf, sz); + if (!p) { + errno = ENOMEM; + goto error; + } + + *buf = p; + *size = sz; + } + + iov.iov_base = *buf; + iov.iov_len = sz; + if (ptrace(PTRACE_GETREGSET, pid, NT_ARM_ZA, &iov)) + goto error; + + za = *buf; + if (za->size <= sz) + break; + + sz = za->size; + } + + return za; + +error: + return NULL; +} + +static int set_za(pid_t pid, const struct user_za_header *za) +{ + struct iovec iov; + + iov.iov_base = (void *)za; + iov.iov_len = za->size; + return ptrace(PTRACE_SETREGSET, pid, NT_ARM_ZA, &iov); +} + +/* Validate attempting to set the specfied VL via ptrace */ +static void ptrace_set_get_vl(pid_t child, unsigned int vl, bool *supported) +{ + struct user_za_header za; + struct user_za_header *new_za = NULL; + size_t new_za_size = 0; + int ret, prctl_vl; + + *supported = false; + + /* Check if the VL is supported in this process */ + prctl_vl = prctl(PR_SME_SET_VL, vl); + if (prctl_vl == -1) + ksft_exit_fail_msg("prctl(PR_SME_SET_VL) failed: %s (%d)\n", + strerror(errno), errno); + + /* If the VL is not supported then a supported VL will be returned */ + *supported = (prctl_vl == vl); + + /* Set the VL by doing a set with no register payload */ + memset(&za, 0, sizeof(za)); + za.size = sizeof(za); + za.vl = vl; + ret = set_za(child, &za); + if (ret != 0) { + ksft_test_result_fail("Failed to set VL %u\n", vl); + return; + } + + /* + * Read back the new register state and verify that we have the + * same VL that we got from prctl() on ourselves. + */ + if (!get_za(child, (void **)&new_za, &new_za_size)) { + ksft_test_result_fail("Failed to read VL %u\n", vl); + return; + } + + ksft_test_result(new_za->vl = prctl_vl, "Set VL %u\n", vl); + + free(new_za); +} + +/* Validate attempting to set no ZA data and read it back */ +static void ptrace_set_no_data(pid_t child, unsigned int vl) +{ + void *read_buf = NULL; + struct user_za_header write_za; + struct user_za_header *read_za; + size_t read_za_size = 0; + int ret; + + /* Set up some data and write it out */ + memset(&write_za, 0, sizeof(write_za)); + write_za.size = ZA_PT_ZA_OFFSET; + write_za.vl = vl; + + ret = set_za(child, &write_za); + if (ret != 0) { + ksft_test_result_fail("Failed to set VL %u no data\n", vl); + return; + } + + /* Read the data back */ + if (!get_za(child, (void **)&read_buf, &read_za_size)) { + ksft_test_result_fail("Failed to read VL %u no data\n", vl); + return; + } + read_za = read_buf; + + /* We might read more data if there's extensions we don't know */ + if (read_za->size < write_za.size) { + ksft_test_result_fail("VL %u wrote %d bytes, only read %d\n", + vl, write_za.size, read_za->size); + goto out_read; + } + + ksft_test_result(read_za->size == write_za.size, + "Disabled ZA for VL %u\n", vl); + +out_read: + free(read_buf); +} + +/* Validate attempting to set data and read it back */ +static void ptrace_set_get_data(pid_t child, unsigned int vl) +{ + void *write_buf; + void *read_buf = NULL; + struct user_za_header *write_za; + struct user_za_header *read_za; + size_t read_za_size = 0; + unsigned int vq = sve_vq_from_vl(vl); + int ret; + size_t data_size; + + data_size = ZA_PT_SIZE(vq); + write_buf = malloc(data_size); + if (!write_buf) { + ksft_test_result_fail("Error allocating %d byte buffer for VL %u\n", + data_size, vl); + return; + } + write_za = write_buf; + + /* Set up some data and write it out */ + memset(write_za, 0, data_size); + write_za->size = data_size; + write_za->vl = vl; + + fill_buf(write_buf + ZA_PT_ZA_OFFSET, ZA_PT_ZA_SIZE(vq)); + + ret = set_za(child, write_za); + if (ret != 0) { + ksft_test_result_fail("Failed to set VL %u data\n", vl); + goto out; + } + + /* Read the data back */ + if (!get_za(child, (void **)&read_buf, &read_za_size)) { + ksft_test_result_fail("Failed to read VL %u data\n", vl); + goto out; + } + read_za = read_buf; + + /* We might read more data if there's extensions we don't know */ + if (read_za->size < write_za->size) { + ksft_test_result_fail("VL %u wrote %d bytes, only read %d\n", + vl, write_za->size, read_za->size); + goto out_read; + } + + ksft_test_result(memcmp(write_buf + ZA_PT_ZA_OFFSET, + read_buf + ZA_PT_ZA_OFFSET, + ZA_PT_ZA_SIZE(vq)) == 0, + "Data match for VL %u\n", vl); + +out_read: + free(read_buf); +out: + free(write_buf); +} + +static int do_parent(pid_t child) +{ + int ret = EXIT_FAILURE; + pid_t pid; + int status; + siginfo_t si; + unsigned int vq, vl; + bool vl_supported; + + /* Attach to the child */ + while (1) { + int sig; + + pid = wait(&status); + if (pid == -1) { + perror("wait"); + goto error; + } + + /* + * This should never happen but it's hard to flag in + * the framework. + */ + if (pid != child) + continue; + + if (WIFEXITED(status) || WIFSIGNALED(status)) + ksft_exit_fail_msg("Child died unexpectedly\n"); + + if (!WIFSTOPPED(status)) + goto error; + + sig = WSTOPSIG(status); + + if (ptrace(PTRACE_GETSIGINFO, pid, NULL, &si)) { + if (errno == ESRCH) + goto disappeared; + + if (errno == EINVAL) { + sig = 0; /* bust group-stop */ + goto cont; + } + + ksft_test_result_fail("PTRACE_GETSIGINFO: %s\n", + strerror(errno)); + goto error; + } + + if (sig == SIGSTOP && si.si_code == SI_TKILL && + si.si_pid == pid) + break; + + cont: + if (ptrace(PTRACE_CONT, pid, NULL, sig)) { + if (errno == ESRCH) + goto disappeared; + + ksft_test_result_fail("PTRACE_CONT: %s\n", + strerror(errno)); + goto error; + } + } + + ksft_print_msg("Parent is %d, child is %d\n", getpid(), child); + + /* Step through every possible VQ */ + for (vq = SVE_VQ_MIN; vq <= SVE_VQ_MAX; vq++) { + vl = sve_vl_from_vq(vq); + + /* First, try to set this vector length */ + ptrace_set_get_vl(child, vl, &vl_supported); + + /* If the VL is supported validate data set/get */ + if (vl_supported) { + ptrace_set_no_data(child, vl); + ptrace_set_get_data(child, vl); + } else { + ksft_test_result_skip("Disabled ZA for VL %u\n", vl); + ksft_test_result_skip("Get and set data for VL %u\n", + vl); + } + } + + ret = EXIT_SUCCESS; + +error: + kill(child, SIGKILL); + +disappeared: + return ret; +} + +int main(void) +{ + int ret = EXIT_SUCCESS; + pid_t child; + + srandom(getpid()); + + ksft_print_header(); + + if (!(getauxval(AT_HWCAP2) & HWCAP2_SME)) { + ksft_set_plan(1); + ksft_exit_skip("SME not available\n"); + } + + ksft_set_plan(EXPECTED_TESTS); + + child = fork(); + if (!child) + return do_child(); + + if (do_parent(child)) + ret = EXIT_FAILURE; + + ksft_print_cnts(); + + return ret; +}

Add a small testcase that attempts to do a clone() with ZA enabled and verifies that it remains enabled with the same contents. We only check one word in one horizontal vector of ZA since there's already other tests that check for data corruption more broadly, we're just looking to make sure that ZA is still enabled and it looks like the data got copied.

Signed-off-by: Mark Brown broonie@kernel.org --- tools/testing/selftests/arm64/fp/.gitignore | 1 + tools/testing/selftests/arm64/fp/Makefile | 9 +- .../testing/selftests/arm64/fp/za-fork-asm.S | 61 +++++++ tools/testing/selftests/arm64/fp/za-fork.c | 156 ++++++++++++++++++ 4 files changed, 226 insertions(+), 1 deletion(-) create mode 100644 tools/testing/selftests/arm64/fp/za-fork-asm.S create mode 100644 tools/testing/selftests/arm64/fp/za-fork.c

diff --git a/tools/testing/selftests/arm64/fp/.gitignore b/tools/testing/selftests/arm64/fp/.gitignore index d98d3d48b504..ea947af63882 100644 --- a/tools/testing/selftests/arm64/fp/.gitignore +++ b/tools/testing/selftests/arm64/fp/.gitignore @@ -8,5 +8,6 @@ sve-test ssve-test vec-syscfg vlset +za-fork za-ptrace za-test diff --git a/tools/testing/selftests/arm64/fp/Makefile b/tools/testing/selftests/arm64/fp/Makefile index 807a8faf8d57..95e707e32247 100644 --- a/tools/testing/selftests/arm64/fp/Makefile +++ b/tools/testing/selftests/arm64/fp/Makefile @@ -1,7 +1,7 @@ # SPDX-License-Identifier: GPL-2.0

CFLAGS += -I../../../../../usr/include/ -TEST_GEN_PROGS := sve-ptrace sve-probe-vls vec-syscfg za-ptrace +TEST_GEN_PROGS := sve-ptrace sve-probe-vls vec-syscfg za-fork za-ptrace TEST_PROGS_EXTENDED := fp-pidbench fpsimd-test fpsimd-stress \ rdvl-sme rdvl-sve \ sve-test sve-stress \ @@ -11,6 +11,7 @@ TEST_PROGS_EXTENDED := fp-pidbench fpsimd-test fpsimd-stress \

all: $(TEST_GEN_PROGS) $(TEST_PROGS_EXTENDED)

+# Build with nolibc to avoid effects due to libc's clone() support fp-pidbench: fp-pidbench.S asm-utils.o $(CC) -nostdlib $^ -o $@ fpsimd-test: fpsimd-test.o asm-utils.o @@ -25,6 +26,12 @@ ssve-test: sve-test.S asm-utils.o $(CC) -DSSVE -nostdlib $^ -o $@ vec-syscfg: vec-syscfg.o rdvl.o vlset: vlset.o +za-fork: za-fork.o za-fork-asm.o + $(CC) -nostdlib -static $^ -o $@ -lgcc +za-fork.o: za-fork.c + $(CC) -c -fno-asynchronous-unwind-tables -fno-ident -s -Os -nostdlib \ + -include ../../../../include/nolibc/nolibc.h \ + -ffreestanding -Wall $^ -o $@ za-test: za-test.o asm-utils.o $(CC) -nostdlib $^ -o $@ za-ptrace: za-ptrace.o diff --git a/tools/testing/selftests/arm64/fp/za-fork-asm.S b/tools/testing/selftests/arm64/fp/za-fork-asm.S new file mode 100644 index 000000000000..947c0962ee1b --- /dev/null +++ b/tools/testing/selftests/arm64/fp/za-fork-asm.S @@ -0,0 +1,61 @@ +// SPDX-License-Identifier: GPL-2.0-only +// Copyright (C) 2021 ARM Limited. + +#include "sme-inst.h" + +.arch_extension sve + +#define MAGIC 42 + +#define MAXVL 2048 +#define MAXVL_B (MAXVL / 8) + +.pushsection .text +.data +.align 4 +scratch: + .space MAXVL_B +.popsection + +.globl fork_test +fork_test: + smstart_za + + // For simplicity just set one word in one vector, other tests + // cover general data corruption issues. + ldr x0, =scratch + mov x1, #MAGIC + str x1, [x0] + mov w12, wzr + _ldr_za 12, 0 // ZA.H[w12] loaded from [x0] + + // Tail call into the C portion that does the fork & verify + b fork_test_c + +.globl verify_fork +verify_fork: + // SVCR should have ZA=1, SM=0 + mrs x0, S3_3_C4_C2_2 + and x1, x0, #3 + cmp x1, #2 + beq 1f + mov x0, xzr + b 100f +1: + + // ZA should still have the value we loaded + ldr x0, =scratch + mov w12, wzr + _str_za 12, 0 // ZA.H[w12] stored to [x0] + ldr x1, [x0] + cmp x1, #MAGIC + beq 2f + mov x0, xzr + b 100f + +2: + // All tests passed + mov x0, #1 +100: + ret + diff --git a/tools/testing/selftests/arm64/fp/za-fork.c b/tools/testing/selftests/arm64/fp/za-fork.c new file mode 100644 index 000000000000..ff475c649e96 --- /dev/null +++ b/tools/testing/selftests/arm64/fp/za-fork.c @@ -0,0 +1,156 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Copyright (C) 2022 ARM Limited. + * Original author: Mark Brown broonie@kernel.org + */ + +// SPDX-License-Identifier: GPL-2.0-only + +#include <linux/sched.h> +#include <linux/wait.h> + +#define EXPECTED_TESTS 1 + +static void putstr(const char *str) +{ + write(1, str, strlen(str)); +} + +static void putnum(unsigned int num) +{ + char c; + + if (num / 10) + putnum(num / 10); + + c = '0' + (num % 10); + write(1, &c, 1); +} + +static int tests_run; +static int tests_passed; +static int tests_failed; +static int tests_skipped; + +static void print_summary(void) +{ + if (tests_passed + tests_failed + tests_skipped != EXPECTED_TESTS) + putstr("# UNEXPECTED TEST COUNT: "); + + putstr("# Totals: pass:"); + putnum(tests_passed); + putstr(" fail:"); + putnum(tests_failed); + putstr(" xfail:0 xpass:0 skip:"); + putnum(tests_skipped); + putstr(" error:0\n"); +} + +int fork_test(void); +int verify_fork(void); + +/* + * If we fork the value in the parent should be unchanged and the + * child should start with the same value. This is called from the + * fork_test() asm function. + */ +int fork_test_c(void) +{ + pid_t newpid, waiting; + int child_status, parent_result; + + newpid = fork(); + if (newpid == 0) { + /* In child */ + if (!verify_fork()) { + putstr("# ZA state invalid in child\n"); + exit(0); + } else { + exit(1); + } + } + if (newpid < 0) { + putstr("# fork() failed: -"); + putnum(-newpid); + putstr("\n"); + return 0; + } + + parent_result = verify_fork(); + if (!parent_result) + putstr("# ZA state invalid in parent\n"); + + for (;;) { + waiting = waitpid(newpid, &child_status, 0); + + if (waiting < 0) { + if (errno == EINTR) + continue; + putstr("# waitpid() failed: "); + putnum(errno); + putstr("\n"); + return 0; + } + if (waiting != newpid) { + putstr("# waitpid() returned wrong PID\n"); + return 0; + } + + if (!WIFEXITED(child_status)) { + putstr("# child did not exit\n"); + return 0; + } + + return WEXITSTATUS(child_status) && parent_result; + } +} + +#define run_test(name) \ + if (name()) { \ + tests_passed++; \ + } else { \ + tests_failed++; \ + putstr("not "); \ + } \ + putstr("ok "); \ + putnum(++tests_run); \ + putstr(" " #name "\n"); + +int main(int argc, char **argv) +{ + int ret, i; + + putstr("TAP version 13\n"); + putstr("1.."); + putnum(EXPECTED_TESTS); + putstr("\n"); + + putstr("# PID: "); + putnum(getpid()); + putstr("\n"); + + /* + * This test is run with nolibc which doesn't support hwcap and + * it's probably disproportionate to implement so instead check + * for the default vector length configuration in /proc. + */ + ret = open("/proc/sys/abi/sme_default_vector_length", O_RDONLY, 0); + if (ret >= 0) { + run_test(fork_test); + + } else { + putstr("# SME support not present\n"); + + for (i = 0; i < EXPECTED_TESTS; i++) { + putstr("ok "); + putnum(i); + putstr(" skipped\n"); + } + + tests_skipped += EXPECTED_TESTS; + } + + print_summary(); + + return 0; +}