December 2024 - Linux-kselftest-mirror

[PATCH net 4/5] selftests: net-drv: queues: sanity check netlink dumps

by Jakub Kicinski

This test already catches a netlink bug fixed by this series, but only when running on HW with many queues. Make sure the netdevsim instance created has a lot of queues, and constrain the size of the recv_buffer used by netlink. While at it test both rx and tx queues. Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> --- CC: shuah(a)kernel.org CC: linux-kselftest(a)vger.kernel.org --- tools/testing/selftests/drivers/net/queues.py | 23 +++++++++++-------- 1 file changed, 13 insertions(+), 10 deletions(-) diff --git a/tools/testing/selftests/drivers/net/queues.py b/tools/testing/selftests/drivers/net/queues.py index 30f29096e27c..9c5473abbd78 100755 --- a/tools/testing/selftests/drivers/net/queues.py +++ b/tools/testing/selftests/drivers/net/queues.py @@ -8,25 +8,28 @@ from lib.py import cmd import glob -def sys_get_queues(ifname) -> int: - folders = glob.glob(f'/sys/class/net/{ifname}/queues/rx-*') +def sys_get_queues(ifname, qtype='rx') -> int: + folders = glob.glob(f'/sys/class/net/{ifname}/queues/{qtype}-*') return len(folders) -def nl_get_queues(cfg, nl): +def nl_get_queues(cfg, nl, qtype='rx'): queues = nl.queue_get({'ifindex': cfg.ifindex}, dump=True) if queues: - return len([q for q in queues if q['type'] == 'rx']) + return len([q for q in queues if q['type'] == qtype]) return None def get_queues(cfg, nl) -> None: - queues = nl_get_queues(cfg, nl) - if not queues: - raise KsftSkipEx('queue-get not supported by device') + snl = NetdevFamily(recv_size=4096) - expected = sys_get_queues(cfg.dev['ifname']) - ksft_eq(queues, expected) + for qtype in ['rx', 'tx']: + queues = nl_get_queues(cfg, snl, qtype) + if not queues: + raise KsftSkipEx('queue-get not supported by device') + + expected = sys_get_queues(cfg.dev['ifname'], qtype) + ksft_eq(queues, expected) def addremove_queues(cfg, nl) -> None: @@ -57,7 +60,7 @@ import glob def main() -> None: - with NetDrvEnv(__file__, queue_count=3) as cfg: + with NetDrvEnv(__file__, queue_count=100) as cfg: ksft_run([get_queues, addremove_queues], args=(cfg, NetdevFamily())) ksft_exit() -- 2.47.1

4 months, 1 week

3
2
0 0

[PATCH net 3/5] selftests: net: support setting recv_size in YNL

by Jakub Kicinski

recv_size parameter allows constraining the buffer size for dumps. It's useful in testing kernel handling of dump continuation, IOW testing dumps which span multiple skbs. Let the tests set this parameter when initializing the YNL family. Keep the normal default, we don't want tests to unintentionally behave very differently than normal code. Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> --- CC: shuah(a)kernel.org CC: jiri(a)resnulli.us CC: petrm(a)nvidia.com CC: linux-kselftest(a)vger.kernel.org --- tools/testing/selftests/net/lib/py/ynl.py | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/tools/testing/selftests/net/lib/py/ynl.py b/tools/testing/selftests/net/lib/py/ynl.py index a0d689d58c57..076a7e8dc3eb 100644 --- a/tools/testing/selftests/net/lib/py/ynl.py +++ b/tools/testing/selftests/net/lib/py/ynl.py @@ -32,23 +32,23 @@ from .ksft import ksft_pr, ktap_result # Set schema='' to avoid jsonschema validation, it's slow # class EthtoolFamily(YnlFamily): - def __init__(self): + def __init__(self, recv_size=0): super().__init__((SPEC_PATH / Path('ethtool.yaml')).as_posix(), - schema='') + schema='', recv_size=recv_size) class RtnlFamily(YnlFamily): - def __init__(self): + def __init__(self, recv_size=0): super().__init__((SPEC_PATH / Path('rt_link.yaml')).as_posix(), - schema='') + schema='', recv_size=recv_size) class NetdevFamily(YnlFamily): - def __init__(self): + def __init__(self, recv_size=0): super().__init__((SPEC_PATH / Path('netdev.yaml')).as_posix(), - schema='') + schema='', recv_size=recv_size) class NetshaperFamily(YnlFamily): - def __init__(self): + def __init__(self, recv_size=0): super().__init__((SPEC_PATH / Path('net_shaper.yaml')).as_posix(), - schema='') + schema='', recv_size=recv_size) -- 2.47.1

4 months, 1 week

3
2
0 0

[PATCH] selftests: arm coresight: sysfsmode testing

by Linu Cherian

Add sysfs mode selftest for ARM Coresight hardware tracer. Signed-off-by: Linu Cherian <lcherian(a)marvell.com> --- .../drivers/hwtracing/coresight/Makefile | 5 + .../hwtracing/coresight/sysfs_test_trace.sh | 144 ++++++++++++++++++ 2 files changed, 149 insertions(+) create mode 100644 tools/testing/selftests/drivers/hwtracing/coresight/Makefile create mode 100755 tools/testing/selftests/drivers/hwtracing/coresight/sysfs_test_trace.sh diff --git a/tools/testing/selftests/drivers/hwtracing/coresight/Makefile b/tools/testing/selftests/drivers/hwtracing/coresight/Makefile new file mode 100644 index 000000000000..7dc68ae1c0a9 --- /dev/null +++ b/tools/testing/selftests/drivers/hwtracing/coresight/Makefile @@ -0,0 +1,5 @@ +# SPDX-License-Identifier: GPL-2.0 + +TEST_PROGS = sysfs_test_trace.sh + +include ../../../lib.mk diff --git a/tools/testing/selftests/drivers/hwtracing/coresight/sysfs_test_trace.sh b/tools/testing/selftests/drivers/hwtracing/coresight/sysfs_test_trace.sh new file mode 100755 index 000000000000..0d6307fff1d2 --- /dev/null +++ b/tools/testing/selftests/drivers/hwtracing/coresight/sysfs_test_trace.sh @@ -0,0 +1,144 @@ +#!/bin/sh +# SPDX-License-Identifier: GPL-2.0 +# Copyright (C) 2024 Marvell. + +# Test Arm CoreSight trace capture in sysfs mode +# Based on tools/perf/tests/shell/test_arm_coresight.sh + +glb_err=0 + +arm_cs_report() { + if [ $2 != 0 ]; then + echo "$1: FAIL" + glb_err=$2 + else + echo "$1: PASS" + fi +} + +is_device_sink() { + # If the node of "enable_sink" is existed under the device path, this + # means the device is a sink device. + + if [ -e "$1/enable_sink" ]; then + + return 0 + else + return 1 + fi +} + +# Configure sink for buffer mode +cfg_sink_buf_mode() { + sink_dev=$1 + mode=$2 + # Set buffer mode if supported + if [ -e "$sink_dev/buf_modes_available" ]; then + cat $sink_dev/buf_modes_available | grep -E -q $mode + if [ $? -eq 0 ]; then + echo $mode > $sink_dev/buf_mode_preferred + return 0 + fi + fi + + return 1 +} + +run_app() { + + taskset -c $1 dd if=/dev/urandom of=/dev/null bs=1M count=64 +} + +sysfs_trace() { + src_dev=$1 + sink_dev=$2 + cpu=$3 + + # Enable sink device + echo 1 > $sink_dev/enable_sink + # Enable source device + echo 1 > $src_dev/enable_source + + # Run app to be traced + run_app $cpu + + # Read back trace data + dd if=/dev/$sink_dev_name of=/tmp/tracedata + + # Verify if read is successful + err=$? + + # Disable source device + echo 0 > $src_dev/enable_source + + # Diskable sink device + echo 0 > $sink_dev/enable_sink + + arm_cs_report "CoreSight path testing (CPU$cpu -> $sink_dev_name)" $err +} + +try_sysfs_trace_resrv_buf() { + src_dev=$1 + sink_dev=$2 + cpu=$3 + + # Configure the resrved buffer mode if available + cfg_sink_buf_mode $sink_dev "resrv" + if [ $? -eq 0 ]; then + echo "Running sysfs trace with resrv buf mode" + sysfs_trace $src_dev $sink_dev $cpu + # Restore buffer mode + cfg_sink_buf_mode $sink_dev "auto" + if [ $? -eq 1 ]; then + echo "Failed to restore default buf mode" + fi + fi +} + +arm_cs_iterate_devices() { + src_dev=$1 + cpu=$3 + for cdev in $2/connections/out\:*; do + + # Skip testing if it's not a directory + ! [ -d $cdev ] && continue; + + # Read out its symbol link file name + sink_dev=`readlink -f $cdev` + + # Extract device name from path, e.g. + # sink_dev = '/sys/devices/platform/20010000.etf/tmc_etf0' + # `> sink_dev_name = 'tmc_etf0' + sink_dev_name=$(basename $sink_dev) + + if is_device_sink $sink_dev; then + # Run trace with resrv buf mode (if available) + try_sysfs_trace_resrv_buf $src_dev $sink_dev $cpu + + # Run the default mode + echo "Running sysfs trace with default settings" + sysfs_trace $src_dev $sink_dev $cpu + fi + + arm_cs_iterate_devices $src_dev $cdev $cpu + + done +} + +arm_cs_etm_traverse_path_test() { + # Iterate for every ETM device + for dev in /sys/bus/event_source/devices/cs_etm/cpu*; do + # Canonicalize the path + dev=`readlink -f $dev` + + # Find the ETM device belonging to which CPU + cpu=`cat $dev/cpu` + + # Use depth-first search (DFS) to iterate outputs + arm_cs_iterate_devices $dev $dev $cpu + done +} + +arm_cs_etm_traverse_path_test + +exit $glb_err -- 2.34.1

4 months, 1 week

3
6
0 0

[PATCH v7] lib/math: Add int_sqrt test suite

by Luis Felipe Hernandez

Adds test suite for integer based square root function. The test suite is designed to verify the correctness of the int_sqrt() math library function. Signed-off-by: Luis Felipe Hernandez <luis.hernandez093(a)gmail.com> --- lib/Kconfig.debug | 15 ++++++++ lib/math/Makefile | 1 + lib/math/tests/Makefile | 1 + lib/math/tests/int_sqrt_kunit.c | 66 +++++++++++++++++++++++++++++++++ 4 files changed, 83 insertions(+) create mode 100644 lib/math/tests/int_sqrt_kunit.c diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug index f3d723705879..147d9fef42e7 100644 --- a/lib/Kconfig.debug +++ b/lib/Kconfig.debug @@ -3161,6 +3161,21 @@ config INT_POW_TEST If unsure, say N +config INT_SQRT_KUNIT_TEST + tristate "Integer square root test" if !KUNIT_ALL_TESTS + depends on KUNIT + default KUNIT_ALL_TESTS + help + This option enables the KUnit test suite for the int_sqrt() function, + which performs square root calculation. The test suite checks + various scenarios, including edge cases, to ensure correctness. + + Enabling this option will include tests that check various scenarios + and edge cases to ensure the accuracy and reliability of the square root + function. + + If unsure, say N + endif # RUNTIME_TESTING_MENU config ARCH_USE_MEMTEST diff --git a/lib/math/Makefile b/lib/math/Makefile index 3ef11305f8d2..853f023ae537 100644 --- a/lib/math/Makefile +++ b/lib/math/Makefile @@ -9,3 +9,4 @@ obj-$(CONFIG_INT_POW_TEST) += tests/int_pow_kunit.o obj-$(CONFIG_TEST_DIV64) += test_div64.o obj-$(CONFIG_TEST_MULDIV64) += test_mul_u64_u64_div_u64.o obj-$(CONFIG_RATIONAL_KUNIT_TEST) += rational-test.o +obj-$(CONFIG_INT_SQRT_KUNIT_TEST) += tests/int_sqrt_kunit.o \ No newline at end of file diff --git a/lib/math/tests/Makefile b/lib/math/tests/Makefile index 6a169123320a..e1a79f093b2d 100644 --- a/lib/math/tests/Makefile +++ b/lib/math/tests/Makefile @@ -1,3 +1,4 @@ # SPDX-License-Identifier: GPL-2.0-only obj-$(CONFIG_INT_POW_TEST) += int_pow_kunit.o +obj-$(CONFIG_INT_SQRT_KUNIT_TEST) += int_sqrt_kunit.o diff --git a/lib/math/tests/int_sqrt_kunit.c b/lib/math/tests/int_sqrt_kunit.c new file mode 100644 index 000000000000..1798e1312eb7 --- /dev/null +++ b/lib/math/tests/int_sqrt_kunit.c @@ -0,0 +1,66 @@ +// SPDX-License-Identifier: GPL-2.0-only + +#include <kunit/test.h> +#include <linux/limits.h> +#include <linux/math.h> +#include <linux/module.h> +#include <linux/string.h> + +struct test_case_params { + unsigned long x; + unsigned long expected_result; + const char *name; +}; + +static const struct test_case_params params[] = { + { 0, 0, "edge case: square root of 0" }, + { 1, 1, "perfect square: square root of 1" }, + { 2, 1, "non-perfect square: square root of 2" }, + { 3, 1, "non-perfect square: square root of 3" }, + { 4, 2, "perfect square: square root of 4" }, + { 5, 2, "non-perfect square: square root of 5" }, + { 6, 2, "non-perfect square: square root of 6" }, + { 7, 2, "non-perfect square: square root of 7" }, + { 8, 2, "non-perfect square: square root of 8" }, + { 9, 3, "perfect square: square root of 9" }, + { 15, 3, "non-perfect square: square root of 15 (N-1 from 16)" }, + { 16, 4, "perfect square: square root of 16" }, + { 17, 4, "non-perfect square: square root of 17 (N+1 from 16)" }, + { 80, 8, "non-perfect square: square root of 80 (N-1 from 81)" }, + { 81, 9, "perfect square: square root of 81" }, + { 82, 9, "non-perfect square: square root of 82 (N+1 from 81)" }, + { 255, 15, "non-perfect square: square root of 255 (N-1 from 256)" }, + { 256, 16, "perfect square: square root of 256" }, + { 257, 16, "non-perfect square: square root of 257 (N+1 from 256)" }, + { 2147483648, 46340, "large input: square root of 2147483648" }, + { 4294967295, 65535, "edge case: ULONG_MAX for 32-bit" }, +}; + +static void get_desc(const struct test_case_params *tc, char *desc) +{ + strscpy(desc, tc->name, KUNIT_PARAM_DESC_SIZE); +} + +KUNIT_ARRAY_PARAM(int_sqrt, params, get_desc); + +static void int_sqrt_test(struct kunit *test) +{ + const struct test_case_params *tc = (const struct test_case_params *)test->param_value; + + KUNIT_EXPECT_EQ(test, tc->expected_result, int_sqrt(tc->x)); +} + +static struct kunit_case math_int_sqrt_test_cases[] = { + KUNIT_CASE_PARAM(int_sqrt_test, int_sqrt_gen_params), + {} +}; + +static struct kunit_suite int_sqrt_test_suite = { + .name = "math-int_sqrt", + .test_cases = math_int_sqrt_test_cases, +}; + +kunit_test_suites(&int_sqrt_test_suite); + +MODULE_DESCRIPTION("math.int_sqrt KUnit test suite"); +MODULE_LICENSE("GPL"); -- 2.47.1

4 months, 1 week

2
1
0 0

📈 Boost Your Rankings in Just 30 Days with Our One-Time SEO Plan!

by SEO MAX

Hi there, That’s right! With our one-time, one-month SEO plan , we’re ready to take your website’s rankings to the next level. Our team will implement proven strategies to optimize your site and secure quality backlinks, all designed to increase your visibility and drive results. Here’s what you’ll get: Full transparency on our work and approach. Comprehensive reports delivered within 30 days. Ready to see the difference? Try us today: https://www.1seo-max.com/get-started/ (https://www.1seo-max.com/get-started/) Or chat with us on Whatsapp: https://www.1seo-max.com/whatsapp-with-us/ (https://www.1seo-max.com/whatsapp-with-us/) Cheers, Mike Arnold Get a free SEO Consultation (https://www.1seo-max.com/free-seo-consultaition/) Unsubscribe (https://clicks.1seo-max.com/?na=u&nk=1182396-88a37ab464&nek=24-) | Manage your subscription (https://clicks.1seo-max.com/?na=p&nk=1182396-88a37ab464&nek=24-) | View online (https://clicks.1seo-max.com/?na=v&nk=1182396-88a37ab464&id=24) whatsapp

4 months, 1 week

1
0
0 0

[PATCH 0/4] KVM: selftests: dirty_log_test: fixes for running the test nested

by Maxim Levitsky

As a part of the effort to start running kvm selftests nested, this patch series contains several fixes to the dirty_log_test, which allows this test to run nested very well. I also included a mostly nop change to KVM, to reverse the order in which the PML log is read to align more closely to the hardware. It should not affect regular users of the dirty logging but it fixes a unit test specific assumption in the dirty_log_test dirty-ring mode. Patch 4 fixes a very rare problem, which is hard to reproduce with standard test parameters, but due to some weird timing issue, it actually happened a few times on my machine which prompted me to investigate it. The issue can be reproduced well by running the test nested (without patch 4 applied) with a very short iteration time and with a few iterations in a loop like this: while ./dirty_log_test -i 10 -I 1 -M dirty-ring ; do true ; done Or even better, it's possible to manually patch the test to not wait at all (effectively setting iteration time to 0), then it fails pretty fast. Best regards, Maxim Levitsky Maxim Levitsky (4): KVM: VMX: read the PML log in the same order as it was written KVM: selftests: dirty_log_test: Limit s390x workaround to s390x KVM: selftests: dirty_log_test: run the guest until some dirty ring entries were harvested KVM: selftests: dirty_log_test: support multiple write retires arch/x86/kvm/vmx/vmx.c | 32 +++++--- arch/x86/kvm/vmx/vmx.h | 1 + tools/testing/selftests/kvm/dirty_log_test.c | 79 +++++++++++++++++--- 3 files changed, 91 insertions(+), 21 deletions(-) -- 2.26.3

4 months, 1 week

2
9
0 0

[PATCH v2 0/3] drivers: base: Don't match device with NULL of_node/fwnode/etc + tests

by Brian Norris

This series: 1. makes the behavior of_find_device_by_node(), bus_find_device_by_of_node(), bus_find_device_by_fwnode(), etc., more consistent when provided with a NULL node/handle; 2. adds kunit tests to validate the new NULL-argument behavior; and 3. makes some related improvements and refactoring for the drivers/base/ kunit tests. This series aims to prevent problems like the ones resolved in commit 5c8418cf4025 ("PCI/pwrctrl: Unregister platform device only if one actually exists"). Changes in v2: * Add Rob's Reviewed-by * CC LKML (oops!) * Keep "devm" and "match" tests in separate suites Brian Norris (3): drivers: base: Don't match devices with NULL of_node/fwnode/etc drivers: base: test: Enable device model tests with KUNIT_ALL_TESTS drivers: base: test: Add ...find_device_by...(... NULL) tests drivers/base/core.c | 8 ++--- drivers/base/test/Kconfig | 1 + drivers/base/test/platform-device-test.c | 42 +++++++++++++++++++++++- 3 files changed, 46 insertions(+), 5 deletions(-) -- 2.47.0.338.g60cca15819-goog

4 months, 1 week

3
6
0 0

[PATCH v2 0/6] Enhance event delivery error handling

by Ivan Orlov

Currently, the situation when guest accesses MMIO during vectoring is handled differently on VMX and SVM: on VMX KVM returns internal error, when SVM goes into infinite loop trying to deliver an event again and again. This patch series eliminates this difference by returning a KVM internal error when guest performs MMIO during vectoring for both VMX and SVM. Also, introduce a selftest test case which covers the error handling mentioned above. V1 -> V2: - Make commit messages more brief, avoid using pronouns - Extract SVM error handling into a separate commit - Introduce a new X86EMUL_ return type and detect the unhandleable vectoring error in vendor-specific check_emulate_instruction instead of handling it in the common MMU code (which is specific for cached MMIO) Ivan Orlov (6): KVM: x86: Add function for vectoring error generation KVM: x86: Add emulation status for vectoring during MMIO KVM: VMX: Handle vectoring error in check_emulate_instruction KVM: SVM: Handle MMIO during vectroing error selftests: KVM: extract lidt into helper function selftests: KVM: Add test case for MMIO during vectoring arch/x86/include/asm/kvm_host.h | 12 ++++- arch/x86/kvm/kvm_emulate.h | 2 + arch/x86/kvm/svm/svm.c | 9 +++- arch/x86/kvm/vmx/vmx.c | 33 +++++------- arch/x86/kvm/x86.c | 27 ++++++++++ .../selftests/kvm/include/x86_64/processor.h | 7 +++ .../selftests/kvm/set_memory_region_test.c | 53 ++++++++++++++++++- .../selftests/kvm/x86_64/sev_smoke_test.c | 2 +- 8 files changed, 119 insertions(+), 26 deletions(-) -- 2.43.0

4 months, 1 week

3
20
0 0

[PATCH v2] kselftest/arm64: abi: fix SVCR detection

by Weizhao Ouyang

When using svcr_in to check ZA and Streaming Mode, we should make sure that the value in x2 is correct, otherwise it may trigger an Illegal instruction if FEAT_SVE and !FEAT_SME. Fixes: 43e3f85523e4 ("kselftest/arm64: Add SME support to syscall ABI test") Signed-off-by: Weizhao Ouyang <o451686892(a)gmail.com> --- .../selftests/arm64/abi/syscall-abi-asm.S | 32 +++++++++---------- 1 file changed, 15 insertions(+), 17 deletions(-) diff --git a/tools/testing/selftests/arm64/abi/syscall-abi-asm.S b/tools/testing/selftests/arm64/abi/syscall-abi-asm.S index df3230fdac39..66ab2e0bae5f 100644 --- a/tools/testing/selftests/arm64/abi/syscall-abi-asm.S +++ b/tools/testing/selftests/arm64/abi/syscall-abi-asm.S @@ -81,32 +81,31 @@ do_syscall: stp x27, x28, [sp, #96] // Set SVCR if we're doing SME - cbz x1, 1f + cbz x1, load_gpr adrp x2, svcr_in ldr x2, [x2, :lo12:svcr_in] msr S3_3_C4_C2_2, x2 -1: // Load ZA and ZT0 if enabled - uses x12 as scratch due to SME LDR - tbz x2, #SVCR_ZA_SHIFT, 1f + tbz x2, #SVCR_ZA_SHIFT, load_gpr mov w12, #0 ldr x2, =za_in -2: _ldr_za 12, 2 +1: _ldr_za 12, 2 add x2, x2, x1 add x12, x12, #1 cmp x1, x12 - bne 2b + bne 1b // ZT0 mrs x2, S3_0_C0_C4_5 // ID_AA64SMFR0_EL1 ubfx x2, x2, #ID_AA64SMFR0_EL1_SMEver_SHIFT, \ #ID_AA64SMFR0_EL1_SMEver_WIDTH - cbz x2, 1f + cbz x2, load_gpr adrp x2, zt_in add x2, x2, :lo12:zt_in _ldr_zt 2 -1: +load_gpr: // Load GPRs x8-x28, and save our SP/FP for later comparison ldr x2, =gpr_in add x2, x2, #64 @@ -125,9 +124,9 @@ do_syscall: str x30, [x2], #8 // LR // Load FPRs if we're not doing neither SVE nor streaming SVE - cbnz x0, 1f + cbnz x0, check_sve_in ldr x2, =svcr_in - tbnz x2, #SVCR_SM_SHIFT, 1f + tbnz x2, #SVCR_SM_SHIFT, check_sve_in ldr x2, =fpr_in ldp q0, q1, [x2] @@ -148,8 +147,8 @@ do_syscall: ldp q30, q31, [x2, #16 * 30] b 2f -1: +check_sve_in: // Load the SVE registers if we're doing SVE/SME ldr x2, =z_in @@ -256,32 +255,31 @@ do_syscall: stp q30, q31, [x2, #16 * 30] // Save SVCR if we're doing SME - cbz x1, 1f + cbz x1, check_sve_out mrs x2, S3_3_C4_C2_2 adrp x3, svcr_out str x2, [x3, :lo12:svcr_out] -1: // Save ZA if it's enabled - uses x12 as scratch due to SME STR - tbz x2, #SVCR_ZA_SHIFT, 1f + tbz x2, #SVCR_ZA_SHIFT, check_sve_out mov w12, #0 ldr x2, =za_out -2: _str_za 12, 2 +1: _str_za 12, 2 add x2, x2, x1 add x12, x12, #1 cmp x1, x12 - bne 2b + bne 1b // ZT0 mrs x2, S3_0_C0_C4_5 // ID_AA64SMFR0_EL1 ubfx x2, x2, #ID_AA64SMFR0_EL1_SMEver_SHIFT, \ #ID_AA64SMFR0_EL1_SMEver_WIDTH - cbz x2, 1f + cbz x2, check_sve_out adrp x2, zt_out add x2, x2, :lo12:zt_out _str_zt 2 -1: +check_sve_out: // Save the SVE state if we have some cbz x0, 1f -- 2.45.2

4 months, 1 week

3
2
0 0

[PATCH v2 0/4] sched: Move task_mm_cid_work to mm delayed work

by Gabriele Monaco

This patchset moves the task_mm_cid_work to a preemptible and migratable context. This reduces the impact of this task to the scheduling latency of real time tasks. The change makes the recurrence of the task a bit more predictable. We also add optimisation and fixes to make sure the task_mm_cid_work works as intended. Patch 1 contains the main changes, removing the task_work on the scheduler tick and using a delayed_work instead. Patch 2 adds some optimisations to the approach, since we rely on a delayed_work, it is no longer required to check that the minimum interval passed since execution, we however terminate the call immediately if we see that no mm_cid is actually active, which could happen on processes sleeping for long time or which exited but whose mm has not been freed yet. Patch 3 allows the mm_cids to be actually compacted when a process reduces its number of threads, which was not the case since the same mm_cids were reused to improve cache locality, more details in [1]. Patch 4 adds a selftest to validate the functionality of the task_mm_cid_work (i.e. to compact the mm_cids), this test requires patch 3 to be applied. Changes since V1 [1]: * Re-arm the delayed_work at each invocation * Cancel the work synchronously at mmdrop * Remove next scan fields and completely rely on the delayed_work * Shrink mm_cid allocation with nr thread/affinity (Mathieu Desnoyers) * Add self test OVERHEAD COMPARISON In this section, I'm going to refer to head as the current state upstream without my patch applied, patch is the same head with these patches applied. Likewise, I'm going to refer to task_mm_cid_work as either the task or the function. The experiments are run on an aarch64 machine with 128 cores. The kernel has a bare configuration with PREEMPT_RT enabled. - Memory The patch introduces some memory overhead: * head uses a callback_head per thread (16 bytes) * patch relies on a delayed work per mm but drops a long (80 bytes net) Tasks with 5 threads or less have lower memory footprint with the current approach. Considering a task_struct can be 7-13 kB and an mm_struct is about 1.4 kB, the overhead should be acceptable. - Boot time I tested the patch booting a virtual machine with vng[2], both head and patch get similar boot times (around 8s). - Runtime I run some rather demanding tests to show what could possibly be a worst case in the approach introduced by this patch. The following tests are running again in vng to have a plain system, running mostly the stressors (if there). Unless differently specified, time is in us. All tests run for 30s. The stress-ng tests were run with 128 stressors, I will omit from the table for clarity. No load head patch running processes(threads): 12(12) 12(12) duration(avg,max,sum): 75,426,987 2,42,45ms invocations: 13 20k stress-ng --cpu-load 80 head patch running processes(threads): 129(129) 129(129) duration(avg,max,sum): 20,2ms,740ms 7,774,280ms invocations: 36k 39k stress-ng --fork head patch running processes(threads): 3.6k(3.6k) 4k(4k) duration(avg,max,sum): 34,41,720 19,457,880ms invocations: 21 46k stress-ng --pthread-max 4 head patch running processes(threads): 129(4k) 129(4k) duration(avg,max,sum): 31,195,41ms 21,1ms,830ms invocations: 1290 38k It is important to note that some of those stressors run for a very short period of time to just fork/create a thread, this heavily favours head since the task won't simply run as often. Moreover, the duration time needs to be read carefully, since the task can now be preempted by threads, I tried to exclude that from the computation, but to keep the probes simple, I didn't exclude interference caused by interrupts. On the same system while isolated, the task runs in about 30-35ms, it is hence highly likely that much larger values are only due to interruptions, rather than the function actually running that long. I will post another email with the scripts used to retrieve the data and more details about the runtime distribution. [1] - https://lore.kernel.org/linux-kernel/20241205083110.180134-2-gmonaco@redhat… [2] - https://github.com/arighi/virtme-ng Gabriele Monaco (3): sched: Move task_mm_cid_work to mm delayed work sched: Remove mm_cid_next_scan as obsolete rseq/selftests: Add test for mm_cid compaction Mathieu Desnoyers (1): sched: Compact RSEQ concurrency IDs with reduced threads and affinity include/linux/mm_types.h | 23 ++- include/linux/sched.h | 1 - kernel/sched/core.c | 66 +------- kernel/sched/sched.h | 32 ++-- tools/testing/selftests/rseq/.gitignore | 1 + tools/testing/selftests/rseq/Makefile | 2 +- .../selftests/rseq/mm_cid_compaction_test.c | 157 ++++++++++++++++++ 7 files changed, 203 insertions(+), 79 deletions(-) create mode 100644 tools/testing/selftests/rseq/mm_cid_compaction_test.c base-commit: 231825b2e1ff6ba799c5eaf396d3ab2354e37c6b -- 2.47.1

4 months, 1 week

2
11
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-kselftest-mirror December 2024