Linux-stable-mirror

linux-stable-mirror@lists.linaro.org

96 participants
123805 discussions

[PATCH v3 0/2] nfc: llcp: fix double put/unlock on LLCP_CLOSED in recv handlers

by Qianchang Zhao

Changes in v3: - Wrap commit messages to <= 75 columns - Use 12-char SHA in Fixes tags - Verify Fixes with git blame (both handlers trace back to d646960f7986) - Run scripts/checkpatch.pl --strict Qianchang Zhao (2): nfc: llcp: avoid double release/put on LLCP_CLOSED in nfc_llcp_recv_disc() nfc: llcp: stop processing on LLCP_CLOSED in nfc_llcp_recv_hdlc() net/nfc/llcp_core.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) -- 2.34.1

1 week, 1 day

[PATCH v2] net: nfc: fix deadlock between nfc_unregister_device and rfkill_fop_write

by Deepanshu Kartikey

A deadlock can occur between nfc_unregister_device() and rfkill_fop_write() due to lock ordering inversion between device_lock and rfkill_global_mutex. The problematic lock order is: Thread A (rfkill_fop_write): rfkill_fop_write() mutex_lock(&rfkill_global_mutex) rfkill_set_block() nfc_rfkill_set_block() nfc_dev_down() device_lock(&dev->dev) <- waits for device_lock Thread B (nfc_unregister_device): nfc_unregister_device() device_lock(&dev->dev) rfkill_unregister() mutex_lock(&rfkill_global_mutex) <- waits for rfkill_global_mutex This creates a classic ABBA deadlock scenario. Fix this by moving rfkill_unregister() and rfkill_destroy() outside the device_lock critical section. Store the rfkill pointer in a local variable before releasing the lock, then call rfkill_unregister() after releasing device_lock. This change is safe because rfkill_fop_write() holds rfkill_global_mutex while calling the rfkill callbacks, and rfkill_unregister() also acquires rfkill_global_mutex before cleanup. Therefore, rfkill_unregister() will wait for any ongoing callback to complete before proceeding, and device_del() is only called after rfkill_unregister() returns, preventing any use-after-free. The similar lock ordering in nfc_register_device() (device_lock -> rfkill_global_mutex via rfkill_register) is safe because during registration the device is not yet in rfkill_list, so no concurrent rfkill operations can occur on this device. Fixes: 3e3b5dfcd16a ("NFC: reorder the logic in nfc_{un,}register_device") Cc: stable(a)vger.kernel.org Reported-by: syzbot+4ef89409a235d804c6c2(a)syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=4ef89409a235d804c6c2 Link: https://lore.kernel.org/all/20251217054908.178907-1-kartikey406@gmail.com/T/ [v1] Signed-off-by: Deepanshu Kartikey <kartikey406(a)gmail.com> --- v2: - Added explanation of why UAF is not possible - Added explanation of why nfc_register_device() is safe - Added Fixes and Cc: stable tags - Fixed blank line after variable declaration (kept it) --- net/nfc/core.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/net/nfc/core.c b/net/nfc/core.c index ae1c842f9c64..82f023f37754 100644 --- a/net/nfc/core.c +++ b/net/nfc/core.c @@ -1154,6 +1154,7 @@ EXPORT_SYMBOL(nfc_register_device); void nfc_unregister_device(struct nfc_dev *dev) { int rc; + struct rfkill *rfk = NULL; pr_debug("dev_name=%s\n", dev_name(&dev->dev)); @@ -1164,13 +1165,17 @@ void nfc_unregister_device(struct nfc_dev *dev) device_lock(&dev->dev); if (dev->rfkill) { - rfkill_unregister(dev->rfkill); - rfkill_destroy(dev->rfkill); + rfk = dev->rfkill; dev->rfkill = NULL; } dev->shutting_down = true; device_unlock(&dev->dev); + if (rfk) { + rfkill_unregister(rfk); + rfkill_destroy(rfk); + } + if (dev->ops->check_presence) { timer_delete_sync(&dev->check_pres_timer); cancel_work_sync(&dev->check_pres_work); -- 2.43.0

1 week, 1 day

[PATCH] arm64: errata: Workaround for SI L1 downstream coherency issue

by Lucas Wei

When software issues a Cache Maintenance Operation (CMO) targeting a dirty cache line, the CPU and DSU cluster may optimize the operation by combining the CopyBack Write and CMO into a single combined CopyBack Write plus CMO transaction presented to the interconnect (MCN). For these combined transactions, the MCN splits the operation into two separate transactions, one Write and one CMO, and then propagates the write and optionally the CMO to the downstream memory system or external Point of Serialization (PoS). However, the MCN may return an early CompCMO response to the DSU cluster before the corresponding Write and CMO transactions have completed at the external PoS or downstream memory. As a result, stale data may be observed by external observers that are directly connected to the external PoS or downstream memory. This erratum affects any system topology in which the following conditions apply: - The Point of Serialization (PoS) is located downstream of the interconnect. - A downstream observer accesses memory directly, bypassing the interconnect. Conditions: This erratum occurs only when all of the following conditions are met: 1. Software executes a data cache maintenance operation, specifically, a clean or invalidate by virtual address (DC CVAC, DC CIVAC, or DC IVAC), that hits on unique dirty data in the CPU or DSU cache. This results in a combined CopyBack and CMO being issued to the interconnect. 2. The interconnect splits the combined transaction into separate Write and CMO transactions and returns an early completion response to the CPU or DSU before the write has completed at the downstream memory or PoS. 3. A downstream observer accesses the affected memory address after the early completion response is issued but before the actual memory write has completed. This allows the observer to read stale data that has not yet been updated at the PoS or downstream memory. The implementation of workaround put a second loop of CMOs at the same virtual address whose operation meet erratum conditions to wait until cache data be cleaned to PoC.. This way of implementation mitigates performance panalty compared to purly duplicate orignial CMO. Cc: stable(a)vger.kernel.org # 6.12.x Signed-off-by: Lucas Wei <lucaswei(a)google.com> --- Documentation/arch/arm64/silicon-errata.rst | 3 ++ arch/arm64/Kconfig | 19 +++++++++++++ arch/arm64/include/asm/assembler.h | 10 +++++++ arch/arm64/kernel/cpu_errata.c | 31 +++++++++++++++++++++ arch/arm64/mm/cache.S | 13 ++++++++- arch/arm64/tools/cpucaps | 1 + 6 files changed, 76 insertions(+), 1 deletion(-) diff --git a/Documentation/arch/arm64/silicon-errata.rst b/Documentation/arch/arm64/silicon-errata.rst index a7ec57060f64..98efdf528719 100644 --- a/Documentation/arch/arm64/silicon-errata.rst +++ b/Documentation/arch/arm64/silicon-errata.rst @@ -213,6 +213,9 @@ stable kernels. | ARM | GIC-700 | #2941627 | ARM64_ERRATUM_2941627 | +----------------+-----------------+-----------------+-----------------------------+ +----------------+-----------------+-----------------+-----------------------------+ +| ARM | SI L1 | #4311569 | ARM64_ERRATUM_4311569 | ++----------------+-----------------+-----------------+-----------------------------+ ++----------------+-----------------+-----------------+-----------------------------+ | Broadcom | Brahma-B53 | N/A | ARM64_ERRATUM_845719 | +----------------+-----------------+-----------------+-----------------------------+ | Broadcom | Brahma-B53 | N/A | ARM64_ERRATUM_843419 | diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig index 93173f0a09c7..89326bb26f48 100644 --- a/arch/arm64/Kconfig +++ b/arch/arm64/Kconfig @@ -1155,6 +1155,25 @@ config ARM64_ERRATUM_3194386 If unsure, say Y. +config ARM64_ERRATUM_4311569 + bool "SI L1: 4311569: workaround for premature CMO completion erratum" + default y + help + This option adds the workaround for ARM SI L1 erratum 4311569. + + The erratum of SI L1 can cause an early response to a combined write + and cache maintenance operation (WR+CMO) before the operation is fully + completed to the Point of Serialization (POS). + This can result in a non-I/O coherent agent observing stale data, + potentially leading to system instability or incorrect behavior. + + Enabling this option implements a software workaround by inserting a + second loop of Cache Maintenance Operation (CMO) immediately following the + end of function to do CMOs. This ensures that the data is correctly serialized + before the buffer is handed off to a non-coherent agent. + + If unsure, say Y. + config CAVIUM_ERRATUM_22375 bool "Cavium erratum 22375, 24313" default y diff --git a/arch/arm64/include/asm/assembler.h b/arch/arm64/include/asm/assembler.h index f0ca7196f6fa..d3d46e5f7188 100644 --- a/arch/arm64/include/asm/assembler.h +++ b/arch/arm64/include/asm/assembler.h @@ -381,6 +381,9 @@ alternative_endif .macro dcache_by_myline_op op, domain, start, end, linesz, tmp, fixup sub \tmp, \linesz, #1 bic \start, \start, \tmp +alternative_if ARM64_WORKAROUND_4311569 + mov \tmp, \start +alternative_else_nop_endif .Ldcache_op\@: .ifc \op, cvau __dcache_op_workaround_clean_cache \op, \start @@ -402,6 +405,13 @@ alternative_endif add \start, \start, \linesz cmp \start, \end b.lo .Ldcache_op\@ +alternative_if ARM64_WORKAROUND_4311569 + .ifnc \op, cvau + mov \start, \tmp + mov \tmp, xzr + cbnz \start, .Ldcache_op\@ + .endif +alternative_else_nop_endif dsb \domain _cond_uaccess_extable .Ldcache_op\@, \fixup diff --git a/arch/arm64/kernel/cpu_errata.c b/arch/arm64/kernel/cpu_errata.c index 8cb3b575a031..c69678c512f1 100644 --- a/arch/arm64/kernel/cpu_errata.c +++ b/arch/arm64/kernel/cpu_errata.c @@ -141,6 +141,30 @@ has_mismatched_cache_type(const struct arm64_cpu_capabilities *entry, return (ctr_real != sys) && (ctr_raw != sys); } +#ifdef CONFIG_ARM64_ERRATUM_4311569 +DEFINE_STATIC_KEY_FALSE(arm_si_l1_workaround_4311569); +static int __init early_arm_si_l1_workaround_4311569_cfg(char *arg) +{ + static_branch_enable(&arm_si_l1_workaround_4311569); + pr_info("Enabling cache maintenance workaround for ARM SI-L1 erratum 4311569\n"); + + return 0; +} +early_param("arm_si_l1_workaround_4311569", early_arm_si_l1_workaround_4311569_cfg); + +/* + * We have some earlier use cases to call cache maintenance operation functions, for example, + * dcache_inval_poc() and dcache_clean_poc() in head.S, before making decision to turn on this + * workaround. Since the scope of this workaround is limited to non-coherent DMA agents, its + * safe to have the workaround off by default. + */ +static bool +need_arm_si_l1_workaround_4311569(const struct arm64_cpu_capabilities *entry, int scope) +{ + return static_branch_unlikely(&arm_si_l1_workaround_4311569); +} +#endif + static void cpu_enable_trap_ctr_access(const struct arm64_cpu_capabilities *cap) { @@ -870,6 +894,13 @@ const struct arm64_cpu_capabilities arm64_errata[] = { ERRATA_MIDR_RANGE_LIST(erratum_spec_ssbs_list), }, #endif +#ifdef CONFIG_ARM64_ERRATUM_4311569 + { + .capability = ARM64_WORKAROUND_4311569, + .type = ARM64_CPUCAP_SYSTEM_FEATURE, + .matches = need_arm_si_l1_workaround_4311569, + }, +#endif #ifdef CONFIG_ARM64_WORKAROUND_SPECULATIVE_UNPRIV_LOAD { .desc = "ARM errata 2966298, 3117295", diff --git a/arch/arm64/mm/cache.S b/arch/arm64/mm/cache.S index 503567c864fd..ddf0097624ed 100644 --- a/arch/arm64/mm/cache.S +++ b/arch/arm64/mm/cache.S @@ -143,9 +143,14 @@ SYM_FUNC_END(dcache_clean_pou) * - end - kernel end address of region */ SYM_FUNC_START(__pi_dcache_inval_poc) +alternative_if ARM64_WORKAROUND_4311569 + mov x4, x0 + mov x5, x1 + mov x6, #1 +alternative_else_nop_endif dcache_line_size x2, x3 sub x3, x2, #1 - tst x1, x3 // end cache line aligned? +again: tst x1, x3 // end cache line aligned? bic x1, x1, x3 b.eq 1f dc civac, x1 // clean & invalidate D / U line @@ -158,6 +163,12 @@ SYM_FUNC_START(__pi_dcache_inval_poc) 3: add x0, x0, x2 cmp x0, x1 b.lo 2b +alternative_if ARM64_WORKAROUND_4311569 + mov x0, x4 + mov x1, x5 + sub x6, x6, #1 + cbz x6, again +alternative_else_nop_endif dsb sy ret SYM_FUNC_END(__pi_dcache_inval_poc) diff --git a/arch/arm64/tools/cpucaps b/arch/arm64/tools/cpucaps index 0fac75f01534..856b6cf6e71e 100644 --- a/arch/arm64/tools/cpucaps +++ b/arch/arm64/tools/cpucaps @@ -103,6 +103,7 @@ WORKAROUND_2077057 WORKAROUND_2457168 WORKAROUND_2645198 WORKAROUND_2658417 +WORKAROUND_4311569 WORKAROUND_AMPERE_AC03_CPU_38 WORKAROUND_AMPERE_AC04_CPU_23 WORKAROUND_TRBE_OVERWRITE_FILL_MODE -- 2.52.0.358.g0dd7633a29-goog

1 week, 2 days

[PATCH] powerpc/kexec: Enable SMT before waking offline CPUs

by Nysal Jan K.A.

If SMT is disabled or a partial SMT state is enabled, when a new kernel image is loaded for kexec, on reboot the following warning is observed: kexec: Waking offline cpu 228. WARNING: CPU: 0 PID: 9062 at arch/powerpc/kexec/core_64.c:223 kexec_prepare_cpus+0x1b0/0x1bc [snip] NIP kexec_prepare_cpus+0x1b0/0x1bc LR kexec_prepare_cpus+0x1a0/0x1bc Call Trace: kexec_prepare_cpus+0x1a0/0x1bc (unreliable) default_machine_kexec+0x160/0x19c machine_kexec+0x80/0x88 kernel_kexec+0xd0/0x118 __do_sys_reboot+0x210/0x2c4 system_call_exception+0x124/0x320 system_call_vectored_common+0x15c/0x2ec This occurs as add_cpu() fails due to cpu_bootable() returning false for CPUs that fail the cpu_smt_thread_allowed() check or non primary threads if SMT is disabled. Fix the issue by enabling SMT and resetting the number of SMT threads to the number of threads per core, before attempting to wake up all present CPUs. Fixes: 38253464bc82 ("cpu/SMT: Create topology_smt_thread_allowed()") Reported-by: Sachin P Bappalige <sachinpb(a)linux.ibm.com> Cc: stable(a)vger.kernel.org # v6.6+ Signed-off-by: Nysal Jan K.A. <nysal(a)linux.ibm.com> --- arch/powerpc/kexec/core_64.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/arch/powerpc/kexec/core_64.c b/arch/powerpc/kexec/core_64.c index 222aa326dace..ff6df43720c4 100644 --- a/arch/powerpc/kexec/core_64.c +++ b/arch/powerpc/kexec/core_64.c @@ -216,6 +216,11 @@ static void wake_offline_cpus(void) { int cpu = 0; + lock_device_hotplug(); + cpu_smt_num_threads = threads_per_core; + cpu_smt_control = CPU_SMT_ENABLED; + unlock_device_hotplug(); + for_each_present_cpu(cpu) { if (!cpu_online(cpu)) { printk(KERN_INFO "kexec: Waking offline cpu %d.\n", -- 2.51.0

1 week, 2 days

[PATCH v2] powerpc: Add reloc_offset() to font bitmap pointer used for bootx_printf()

by Finn Thain

Since Linux v6.7, booting using BootX on an Old World PowerMac produces an early crash. Stan Johnson writes, "the symptoms are that the screen goes blank and the backlight stays on, and the system freezes (Linux doesn't boot)." Further testing revealed that the failure can be avoided by disabling CONFIG_BOOTX_TEXT. Bisection revealed that the regression was caused by a change to the font bitmap pointer that's used when btext_init() begins painting characters on the display, early in the boot process. Christophe Leroy explains, "before kernel text is relocated to its final location ... data is addressed with an offset which is added to the Global Offset Table (GOT) entries at the start of bootx_init() by function reloc_got2(). But the pointers that are located inside a structure are not referenced in the GOT and are therefore not updated by reloc_got2(). It is therefore needed to apply the offset manually by using PTRRELOC() macro." Cc: Cedar Maxwell <cedarmaxwell(a)mac.com> Cc: Stan Johnson <userm57(a)yahoo.com> Cc: "Dr. David Alan Gilbert" <linux(a)treblig.org> Cc: Benjamin Herrenschmidt <benh(a)kernel.crashing.org> Cc: stable(a)vger.kernel.org Link: https://lists.debian.org/debian-powerpc/2025/10/msg00111.html Link: https://lore.kernel.org/linuxppc-dev/d81ddca8-c5ee-d583-d579-02b19ed95301@y… Reported-by: Cedar Maxwell <cedarmaxwell(a)mac.com> Closes: https://lists.debian.org/debian-powerpc/2025/09/msg00031.html Bisected-by: Stan Johnson <userm57(a)yahoo.com> Tested-by: Stan Johnson <userm57(a)yahoo.com> Fixes: 0ebc7feae79a ("powerpc: Use shared font data") Suggested-by: Christophe Leroy <christophe.leroy(a)csgroup.eu> Signed-off-by: Finn Thain <fthain(a)linux-m68k.org> --- Changed since v1: - Improved commit log entry to better explain the need for PTRRELOC(). --- arch/powerpc/kernel/btext.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/arch/powerpc/kernel/btext.c b/arch/powerpc/kernel/btext.c index 7f63f1cdc6c3..ca00c4824e31 100644 --- a/arch/powerpc/kernel/btext.c +++ b/arch/powerpc/kernel/btext.c @@ -20,6 +20,7 @@ #include <asm/io.h> #include <asm/processor.h> #include <asm/udbg.h> +#include <asm/setup.h> #define NO_SCROLL @@ -463,7 +464,7 @@ static noinline void draw_byte(unsigned char c, long locX, long locY) { unsigned char *base = calc_base(locX << 3, locY << 4); unsigned int font_index = c * 16; - const unsigned char *font = font_sun_8x16.data + font_index; + const unsigned char *font = PTRRELOC(font_sun_8x16.data) + font_index; int rb = dispDeviceRowBytes; rmci_maybe_on(); -- 2.49.1

1 week, 2 days

Re: [PATCH v2] fbdev: bitblit: bound-check glyph index in bit_putcs*

by Vitaly Chikunov

Dear linux-fbdev, stable, On Mon, Oct 20, 2025 at 09:47:01PM +0800, Junjie Cao wrote: > bit_putcs_aligned()/unaligned() derived the glyph pointer from the > character value masked by 0xff/0x1ff, which may exceed the actual font's > glyph count and read past the end of the built-in font array. > Clamp the index to the actual glyph count before computing the address. > > This fixes a global out-of-bounds read reported by syzbot. > > Reported-by: syzbot+793cf822d213be1a74f2(a)syzkaller.appspotmail.com > Closes: https://syzkaller.appspot.com/bug?extid=793cf822d213be1a74f2 > Tested-by: syzbot+793cf822d213be1a74f2(a)syzkaller.appspotmail.com > Signed-off-by: Junjie Cao <junjie.cao(a)intel.com> This commit is applied to v5.10.247 and causes a regression: when switching VT with ctrl-alt-f2 the screen is blank or completely filled with angle characters, then new text is not appearing (or not visible). This commit is found with git bisect from v5.10.246 to v5.10.247: 0998a6cb232674408a03e8561dc15aa266b2f53b is the first bad commit commit 0998a6cb232674408a03e8561dc15aa266b2f53b Author: Junjie Cao <junjie.cao(a)intel.com> AuthorDate: 2025-10-20 21:47:01 +0800 Commit: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> CommitDate: 2025-12-07 06:08:07 +0900 fbdev: bitblit: bound-check glyph index in bit_putcs* commit 18c4ef4e765a798b47980555ed665d78b71aeadf upstream. bit_putcs_aligned()/unaligned() derived the glyph pointer from the character value masked by 0xff/0x1ff, which may exceed the actual font's glyph count and read past the end of the built-in font array. Clamp the index to the actual glyph count before computing the address. This fixes a global out-of-bounds read reported by syzbot. Reported-by: syzbot+793cf822d213be1a74f2(a)syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=793cf822d213be1a74f2 Tested-by: syzbot+793cf822d213be1a74f2(a)syzkaller.appspotmail.com Signed-off-by: Junjie Cao <junjie.cao(a)intel.com> Reviewed-by: Thomas Zimmermann <tzimmermann(a)suse.de> Signed-off-by: Helge Deller <deller(a)gmx.de> Cc: stable(a)vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> drivers/video/fbdev/core/bitblit.c | 16 ++++++++++++---- 1 file changed, 12 insertions(+), 4 deletions(-) The minimal reproducer in cli, after kernel is booted: date >/dev/tty2; chvt 2 and the date does not appear. Thanks, #regzbot introduced: 0998a6cb232674408a03e8561dc15aa266b2f53b > --- > v1: https://lore.kernel.org/linux-fbdev/5d237d1a-a528-4205-a4d8-71709134f1e1@su… > v1 -> v2: > - Fix indentation and add blank line after declarations with the .pl helper > - No functional changes > > drivers/video/fbdev/core/bitblit.c | 16 ++++++++++++---- > 1 file changed, 12 insertions(+), 4 deletions(-) > > diff --git a/drivers/video/fbdev/core/bitblit.c b/drivers/video/fbdev/core/bitblit.c > index 9d2e59796c3e..085ffb44c51a 100644 > --- a/drivers/video/fbdev/core/bitblit.c > +++ b/drivers/video/fbdev/core/bitblit.c > @@ -79,12 +79,16 @@ static inline void bit_putcs_aligned(struct vc_data *vc, struct fb_info *info, > struct fb_image *image, u8 *buf, u8 *dst) > { > u16 charmask = vc->vc_hi_font_mask ? 0x1ff : 0xff; > + unsigned int charcnt = vc->vc_font.charcount; > u32 idx = vc->vc_font.width >> 3; > u8 *src; > > while (cnt--) { > - src = vc->vc_font.data + (scr_readw(s++)& > - charmask)*cellsize; > + u16 ch = scr_readw(s++) & charmask; > + > + if (ch >= charcnt) > + ch = 0; > + src = vc->vc_font.data + (unsigned int)ch * cellsize; > > if (attr) { > update_attr(buf, src, attr, vc); > @@ -112,14 +116,18 @@ static inline void bit_putcs_unaligned(struct vc_data *vc, > u8 *dst) > { > u16 charmask = vc->vc_hi_font_mask ? 0x1ff : 0xff; > + unsigned int charcnt = vc->vc_font.charcount; > u32 shift_low = 0, mod = vc->vc_font.width % 8; > u32 shift_high = 8; > u32 idx = vc->vc_font.width >> 3; > u8 *src; > > while (cnt--) { > - src = vc->vc_font.data + (scr_readw(s++)& > - charmask)*cellsize; > + u16 ch = scr_readw(s++) & charmask; > + > + if (ch >= charcnt) > + ch = 0; > + src = vc->vc_font.data + (unsigned int)ch * cellsize; > > if (attr) { > update_attr(buf, src, attr, vc); > -- > 2.48.1 >

1 week, 2 days

[PATCH] SUNRPC: auth_gss: fix memory leaks in XDR decoding error paths

by Chuck Lever

From: Chuck Lever <chuck.lever(a)oracle.com> The gssx_dec_ctx(), gssx_dec_status(), and gssx_dec_name() functions allocate memory via gssx_dec_buffer(), which calls kmemdup(). When a subsequent decode operation fails, these functions return immediately without freeing previously allocated buffers, causing memory leaks. The leak in gssx_dec_ctx() is particularly relevant because the caller (gssp_accept_sec_context_upcall) initializes several buffer length fields to non-zero values, resulting in memory allocation: struct gssx_ctx rctxh = { .exported_context_token.len = GSSX_max_output_handle_sz, .mech.len = GSS_OID_MAX_LEN, .src_name.display_name.len = GSSX_max_princ_sz, .targ_name.display_name.len = GSSX_max_princ_sz }; If, for example, gssx_dec_name() succeeds for src_name but fails for targ_name, the memory allocated for exported_context_token, mech, and src_name.display_name remains unreferenced and cannot be reclaimed. Add error handling with goto-based cleanup to free any previously allocated buffers before returning an error. Reported-by: Xingjing Deng <micro6947(a)gmail.com> Closes: https://lore.kernel.org/linux-nfs/CAK+ZN9qttsFDu6h1FoqGadXjMx1QXqPMoYQ=6O9R… Fixes: 1d658336b05f ("SUNRPC: Add RPC based upcall mechanism for RPCGSS auth") Cc: stable(a)vger.kernel.org Signed-off-by: Chuck Lever <chuck.lever(a)oracle.com> --- net/sunrpc/auth_gss/gss_rpc_xdr.c | 82 ++++++++++++++++++++++++------- 1 file changed, 64 insertions(+), 18 deletions(-) diff --git a/net/sunrpc/auth_gss/gss_rpc_xdr.c b/net/sunrpc/auth_gss/gss_rpc_xdr.c index 7d2cdc2bd374..f320c0a8e604 100644 --- a/net/sunrpc/auth_gss/gss_rpc_xdr.c +++ b/net/sunrpc/auth_gss/gss_rpc_xdr.c @@ -320,29 +320,47 @@ static int gssx_dec_status(struct xdr_stream *xdr, /* status->minor_status */ p = xdr_inline_decode(xdr, 8); - if (unlikely(p == NULL)) - return -ENOSPC; + if (unlikely(p == NULL)) { + err = -ENOSPC; + goto out_free_mech; + } p = xdr_decode_hyper(p, &status->minor_status); /* status->major_status_string */ err = gssx_dec_buffer(xdr, &status->major_status_string); if (err) - return err; + goto out_free_mech; /* status->minor_status_string */ err = gssx_dec_buffer(xdr, &status->minor_status_string); if (err) - return err; + goto out_free_major_status_string; /* status->server_ctx */ err = gssx_dec_buffer(xdr, &status->server_ctx); if (err) - return err; + goto out_free_minor_status_string; /* we assume we have no options for now, so simply consume them */ /* status->options */ err = dummy_dec_opt_array(xdr, &status->options); + if (err) + goto out_free_server_ctx; + return 0; + +out_free_server_ctx: + kfree(status->server_ctx.data); + status->server_ctx.data = NULL; +out_free_minor_status_string: + kfree(status->minor_status_string.data); + status->minor_status_string.data = NULL; +out_free_major_status_string: + kfree(status->major_status_string.data); + status->major_status_string.data = NULL; +out_free_mech: + kfree(status->mech.data); + status->mech.data = NULL; return err; } @@ -505,28 +523,35 @@ static int gssx_dec_name(struct xdr_stream *xdr, /* name->name_type */ err = gssx_dec_buffer(xdr, &dummy_netobj); if (err) - return err; + goto out_free_display_name; /* name->exported_name */ err = gssx_dec_buffer(xdr, &dummy_netobj); if (err) - return err; + goto out_free_display_name; /* name->exported_composite_name */ err = gssx_dec_buffer(xdr, &dummy_netobj); if (err) - return err; + goto out_free_display_name; /* we assume we have no attributes for now, so simply consume them */ /* name->name_attributes */ err = dummy_dec_nameattr_array(xdr, &dummy_name_attr_array); if (err) - return err; + goto out_free_display_name; /* we assume we have no options for now, so simply consume them */ /* name->extensions */ err = dummy_dec_opt_array(xdr, &dummy_option_array); + if (err) + goto out_free_display_name; + return 0; + +out_free_display_name: + kfree(name->display_name.data); + name->display_name.data = NULL; return err; } @@ -649,32 +674,34 @@ static int gssx_dec_ctx(struct xdr_stream *xdr, /* ctx->state */ err = gssx_dec_buffer(xdr, &ctx->state); if (err) - return err; + goto out_free_exported_context_token; /* ctx->need_release */ err = gssx_dec_bool(xdr, &ctx->need_release); if (err) - return err; + goto out_free_state; /* ctx->mech */ err = gssx_dec_buffer(xdr, &ctx->mech); if (err) - return err; + goto out_free_state; /* ctx->src_name */ err = gssx_dec_name(xdr, &ctx->src_name); if (err) - return err; + goto out_free_mech; /* ctx->targ_name */ err = gssx_dec_name(xdr, &ctx->targ_name); if (err) - return err; + goto out_free_src_name; /* ctx->lifetime */ p = xdr_inline_decode(xdr, 8+8); - if (unlikely(p == NULL)) - return -ENOSPC; + if (unlikely(p == NULL)) { + err = -ENOSPC; + goto out_free_targ_name; + } p = xdr_decode_hyper(p, &ctx->lifetime); /* ctx->ctx_flags */ @@ -683,17 +710,36 @@ static int gssx_dec_ctx(struct xdr_stream *xdr, /* ctx->locally_initiated */ err = gssx_dec_bool(xdr, &ctx->locally_initiated); if (err) - return err; + goto out_free_targ_name; /* ctx->open */ err = gssx_dec_bool(xdr, &ctx->open); if (err) - return err; + goto out_free_targ_name; /* we assume we have no options for now, so simply consume them */ /* ctx->options */ err = dummy_dec_opt_array(xdr, &ctx->options); + if (err) + goto out_free_targ_name; + return 0; + +out_free_targ_name: + kfree(ctx->targ_name.display_name.data); + ctx->targ_name.display_name.data = NULL; +out_free_src_name: + kfree(ctx->src_name.display_name.data); + ctx->src_name.display_name.data = NULL; +out_free_mech: + kfree(ctx->mech.data); + ctx->mech.data = NULL; +out_free_state: + kfree(ctx->state.data); + ctx->state.data = NULL; +out_free_exported_context_token: + kfree(ctx->exported_context_token.data); + ctx->exported_context_token.data = NULL; return err; } -- 2.52.0

1 week, 2 days

Re: [REGRESSION] erofs: new file-backed stacking limit breaks container overlay use case in 6.12.63

by Gao Xiang

Hi Alekséi, On 2025/12/26 20:17, Alekséi Naidénov wrote: > Hello, > > I am reporting a regression in the 6.12 stable series related to EROFS > file-backed mounts. > > After updating from Linux 6.12.62 to 6.12.63, a previously working setup > using OSTree-backed composefs mounts as Podman rootfs no longer works. > > The regression appears to be caused by the following commit: > > 34447aeedbaea8f9aad3da5b07030a1c0e124639 ("erofs: limit the level of fs > stacking for file-backed mounts") > (backport of upstream commit d53cd891f0e4311889349fff3a784dc552f814b9) > > ## Setup description > > We use OSTree to materialize filesystem trees, which are mounted via > composefs (EROFS + overlayfs) as a read-only filesystem. This mounted > composefs tree is then used as a Podman rootfs, with Podman mounting a > writable overlayfs on top for each container. > > This setup worked correctly on Linux 6.12.62 and earlier. The following issue just tracks this: https://github.com/coreos/fedora-coreos-tracker/issues/2087 I don't think more information is needed, but I really think the EROFS commit is needed to avoid kernel stack overflow due to nested fses. > > In short, the stacking looks like: > > EROFS (file-backed) > -> composefs (EROFS + overlayfs with ostree repo as datadir, read-only) > -> Podman rootfs overlays (RW upperdir) > > There is no recursive or self-stacking of EROFS. Yes, but there are two overlayfs + one file-backed EROFS already, and it exceeds FILESYSTEM_MAX_STACK_DEPTH. That is overlayfs refuses to mount the nested fses. Thanks, Gao Xiang

1 week, 2 days

[REGRESSION] erofs: new file-backed stacking limit breaks container overlay use case in 6.12.63

by Alekséi Naidénov

Hello, I am reporting a regression in the 6.12 stable series related to EROFS file-backed mounts. After updating from Linux 6.12.62 to 6.12.63, a previously working setup using OSTree-backed composefs mounts as Podman rootfs no longer works. The regression appears to be caused by the following commit: 34447aeedbaea8f9aad3da5b07030a1c0e124639 ("erofs: limit the level of fs stacking for file-backed mounts") (backport of upstream commit d53cd891f0e4311889349fff3a784dc552f814b9) ## Setup description We use OSTree to materialize filesystem trees, which are mounted via composefs (EROFS + overlayfs) as a read-only filesystem. This mounted composefs tree is then used as a Podman rootfs, with Podman mounting a writable overlayfs on top for each container. This setup worked correctly on Linux 6.12.62 and earlier. In short, the stacking looks like: EROFS (file-backed) -> composefs (EROFS + overlayfs with ostree repo as datadir, read-only) -> Podman rootfs overlays (RW upperdir) There is no recursive or self-stacking of EROFS. ## Working case (6.12.62) The composefs mount exists and Podman can successfully start a container using it as rootfs. Example composefs mount: ❯ mount | grep a31550cc69eef0e3227fa700623250592711fdfd51b5403a74288b55e89e7e8c a31550cc69eef0e3227fa700623250592711fdfd51b5403a74288b55e89e7e8c on /home/growler/.local/share/containers/ostree/a31550cc69eef0e3227fa700623250592711fdfd51b5403a74288b55e89e7e8c type overlay (ro,noatime,lowerdir+=/proc/self/fd/10,datadir+=/proc/self/fd/7,redirect_dir=on,metacopy=on) (lowedir is a handle for the erofs file-backed mount, datadir is a handle for the ostree repository objects directory) Running Podman: ❯ podman run --rm -it --rootfs $HOME/.local/share/containers/ostree/a31550cc69eef0e3227fa700623250592711fdfd51b5403a74288b55e89e7e8c:O bash -l root@d691e785bba3:/# uname -a Linux d691e785bba3 6.12.62 #1-NixOS SMP PREEMPT_DYNAMIC Fri Dec 12 17:37:22 UTC 2025 x86_64 GNU/Linux root@d691e785bba3:/# (succeed) ## Failing case (6.12.63) After upgrading to 6.12.63, the same command fails when Podman tries to create the writable overlay on top of the composefs mount. Error: ❯ podman run --rm -it --rootfs $HOME/.local/share/containers/ostree/a31550cc69eef0e3227fa700623250592711fdfd51b5403a74288b55e89e7e8c:O bash -l Error: rootfs-overlay: creating overlay failed "/home/growler/.local/share/containers/ostree/a31550cc69eef0e3227fa700623250592711fdfd51b5403a74288b55e89e7e8c" from native overlay: mount overlay:/home/growler/.local/share/containers/storage/overlay-containers/a0851294d6b5b18062d4f5316032ee84d7bae700ea7d12c5be949d9e1999b0a1/rootfs/merge, flags: 0x4, data: lowerdir=/home/growler/.local/share/containers/ostree/a31550cc69eef0e3227fa700623250592711fdfd51b5403a74288b55e89e7e8c,upperdir=/home/growler/.local/share/containers/storage/overlay-containers/a0851294d6b5b18062d4f5316032ee84d7bae700ea7d12c5be949d9e1999b0a1/rootfs/upper,workdir=/home/growler/.local/share/containers/storage/overlay-containers/a0851294d6b5b18062d4f5316032ee84d7bae700ea7d12c5be949d9e1999b0a1/rootfs/work,userxattr: invalid argument ❯ uname -a Linux ci-node-09 6.12.63 #1-NixOS SMP PREEMPT_DYNAMIC Thu Dec 18 12:55:23 UTC 2025 x86_64 GNU/Linux ## Expected behavior Using a composefs (EROFS + overlayfs) read-only mount as the lowerdir for a container rootfs overlay should continue to work as it did in 6.12.62. ## Actual behavior Overlayfs mounting fails with EINVAL when stacking on top of the composefs mount backed by EROFS. ## Notes The setup does not involve recursive EROFS mounting or unbounded stacking depth. It appears the new stacking limit rejects this valid and previously supported container use case. Please let me know if further details or testing would be helpful. Thank you, -- Alekséi Nadénov

1 week, 2 days

[PATCH 6.6.y 0/2] fix push_nsh() validation + silence selftest warnings

by Adrian Yip

Hi maintainers, This is a v6.6 backport mainly for an upstream commit `5ace7ef87f05 net: openvswitch: fix middle attribute validation in push_nsh() action`. I built the kernel then tested it with selftest. The selftest that ran with a a bunch of SyntaxWarning. Example: /ovs-dpctl.py:598: SyntaxWarning: invalid escape sequence '\d' actstr, ":", "(\d+)", int, False /ovs-dpctl.py:601: SyntaxWarning: invalid escape sequence '\d' actstr, "-", "(\d+)", int, False /ovs-dpctl.py:505: SyntaxWarning: invalid escape sequence '\d' elif parse_starts_block(actstr, "^(\d+)", False, True): This error was then easily fixed with another minimal backport for the file tools/testing/selftests/net/openvswitch/ovs-dpctl.py. Hence the series. Both patches was applied cleanly and was tested with selftest and passed though the timeout had to be increased for drop_reason to pass. Adrian Moreno (1): selftests: openvswitch: Fix escape chars in regexp. Ilya Maximets (1): net: openvswitch: fix middle attribute validation in push_nsh() action net/openvswitch/flow_netlink.c | 13 ++++++++++--- .../selftests/net/openvswitch/ovs-dpctl.py | 16 ++++++++-------- 2 files changed, 18 insertions(+), 11 deletions(-) -- 2.52.0

1 week, 3 days

Jump to page:

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror