{"api_version":"1","generated_at":"2026-10-06T10:21:34+00:00","cve":"CVE-2026-98256","urls":{"html":"https://cve.report/CVE-2026-98256","api":"https://cve.report/api/cve/CVE-2026-98256.json","docs":"https://cve.report/api","cve_org":"https://www.cve.org/CVERecord?id=CVE-2026-98256","nvd":"https://nvd.nist.gov/vuln/detail/CVE-2026-98256"},"summary":{"title":"signal: Prevent exec() race","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nsignal: Prevent exec() race\n\nHyunwoo debugged the following KASAN UAF splat:\n\n  BUG: KASAN: slab-use-after-free in __send_signal_locked+0xb27/0xba0\n  Write of size 8 at addr ffff888007ed80c8 by task poc/79\n  ...\n  Call Trace:\n   __send_signal_locked+0xb27/0xba0\n   do_send_sig_info+0xa7/0x160\n   do_send_specific+0x76/0xa0\n   __x64_sys_tgkill+0x193/0x270\n  ...\n  Allocated by task 80:\n   do_timer_create+0x1a4/0x1030\n   __x64_sys_timer_create+0x145/0x190\n  ...\n  Freed by task 12:\n   kmem_cache_free_bulk+0x1f8/0x4a0\n   kvfree_rcu_bulk+0x14f/0x1c0\n   kfree_rcu_work+0x128/0x1a0\n  ...\n  Last potentially related work creation:\n   kvfree_call_rcu+0x39/0x390\n   __flush_itimer_signals+0x211/0x320\n   flush_itimer_signals+0x47/0x90\n   begin_new_exec+0xa6b/0x28c0\n\nIt turned out that this happens with a non-leader exec() as Hyunwoo\nexplained:\n\nde_thread() calls exchange_tids() before release_task(leader), so the\nstruct pid held by a SIGEV_THREAD_ID timer created against the leader's tid\nnow points to the thread which called execve(). pid_task() returns that\nthread and lock_task_sighand() on it succeeds.\n\nIf the timer signal is blocked, its sigqueue stays queued on the leader's\ntask::pending. The next expiry of that timer can then run while\nrelease_task() flushes the queue.\n\nposixtimer_send_sigqueue() checks whether the sigqueue is already queued\nwith a plain list_empty(), which only reads list_head::next.\nlist_del_init() is not atomic and INIT_LIST_HEAD() stores list_head::next\nbefore list_head::prev, so the check can pass in between. list_add_tail()\nqueues the entry on the task::pending of the live thread, and the\nlist_head::prev store from the flush then overwrites the list_head::prev\nlink that list_add_tail() has just set.\n\n__flush_itimer_signals() does not undo that either. With list_head::prev\npointing at the entry itself, its list_del_init() only stores the same\nvalues again, so the entry is not removed from the list. It is still there\nafter the last reference is dropped and the timer is freed by RCU, and the\nlist_add_tail() of a later tgkill() follows that list_head::prev into the\nfreed timer.\n\nThis problem surfaced with the recent commit which moved the sigqueue flush\nout of the sighand lock held region.\n\nHyonwoo proposed to fix this by using list_del_init_careful(), but that\njust papers over the problem. After some disucssions and various attempts\nto solve it, Eric pointed out that there is no reason to flush\ntask::pending late in release_task() and it should be done in\nexit_signals() already.\n\nAs nothing can collect and deliver signals which are queued in a dying\ntask's pending queue, there is no reason to delay it further.\n\nBut it has to be ensured that no signals can be queued into it after that\npoint. exit_signals() sets PF_EXITING in task::flags, which can be used as\nan indicator for this.\n\nCure it by:\n\n  - Preventing signal queueing for task private signals (PIDTYPE_PID) when\n    the task has PF_EXITING set in __send_signal_locked() and in\n    posixtimer_send_sigqueue().\n\n  - Protecting the unlocked setting of PF_EXITING in exit_signals() for the\n    task group empty and the group exit case with sighand lock\n\n  - Flushing task::pending signals right there.\n\n    Optimize that by moving the whole pending list to an on-stack list head\n    under sighand lock and free the signals without the lock held.\n\nThere has been quite some discussion about the lockless flush and the\nnon-leader exec case on weakly ordered systems. The problem is that a third\nparty which tries to send a posix timer signal relies on the PID lookup to\nfind the target task and that lookup might result in the new leader when\nthe signal was originaly directed to the old leader. In case that the\nsignal was queued on the old leader then the lockless flush raised a\nconcern over the following situation:\n\n   old_leader\t\tnew_leader              third party\n\nA: flush_list()\t// list_del_in\n---truncated---","state":"PUBLISHED","assigner":"Linux","published_at":"2026-10-06 09:18:14","updated_at":"2026-10-06 09:18:14"},"problem_types":[],"metrics":[],"references":[{"url":"https://git.kernel.org/stable/c/621c365739828900f32a0c1cdfe5387528a62176","name":"https://git.kernel.org/stable/c/621c365739828900f32a0c1cdfe5387528a62176","refsource":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","tags":[],"title":"","mime":"","httpstatus":"","archivestatus":"0"},{"url":"https://git.kernel.org/stable/c/d2710c8d938ae6a825a6463158e6e6f31eac792a","name":"https://git.kernel.org/stable/c/d2710c8d938ae6a825a6463158e6e6f31eac792a","refsource":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","tags":[],"title":"","mime":"","httpstatus":"","archivestatus":"0"},{"url":"https://git.kernel.org/stable/c/934bd95f1a6404dbb845951b823547abab05a649","name":"https://git.kernel.org/stable/c/934bd95f1a6404dbb845951b823547abab05a649","refsource":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","tags":[],"title":"","mime":"","httpstatus":"","archivestatus":"0"},{"url":"https://www.cve.org/CVERecord?id=CVE-2026-98256","name":"CVE Program record","refsource":"CVE.ORG","tags":["canonical"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-98256","name":"NVD vulnerability detail","refsource":"NVD","tags":["canonical","analysis"]}],"affected":[{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected fb3bbcfe344e64a46574a638b051ffd78762c12d 621c365739828900f32a0c1cdfe5387528a62176 git","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected fb3bbcfe344e64a46574a638b051ffd78762c12d 934bd95f1a6404dbb845951b823547abab05a649 git","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected fb3bbcfe344e64a46574a638b051ffd78762c12d d2710c8d938ae6a825a6463158e6e6f31eac792a git","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected 6.15","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"unaffected 6.15 semver","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"unaffected 6.18.54 6.18.* semver","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"unaffected 7.2.8 7.2.* semver","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"unaffected 7.3-rc4 * original_commit_for_fix","platforms":[]}],"timeline":[],"solutions":[],"workarounds":[],"exploits":[],"credits":[],"nvd_cpes":[],"vendor_comments":[],"enrichments":{"kev":null,"epss":null,"legacy_qids":[]},"source_records":{"cve_program":{"containers":{"cna":{"affected":[{"defaultStatus":"unaffected","product":"Linux","programFiles":["kernel/exit.c","kernel/signal.c"],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","vendor":"Linux","versions":[{"lessThan":"621c365739828900f32a0c1cdfe5387528a62176","status":"affected","version":"fb3bbcfe344e64a46574a638b051ffd78762c12d","versionType":"git"},{"lessThan":"934bd95f1a6404dbb845951b823547abab05a649","status":"affected","version":"fb3bbcfe344e64a46574a638b051ffd78762c12d","versionType":"git"},{"lessThan":"d2710c8d938ae6a825a6463158e6e6f31eac792a","status":"affected","version":"fb3bbcfe344e64a46574a638b051ffd78762c12d","versionType":"git"}]},{"defaultStatus":"affected","product":"Linux","programFiles":["kernel/exit.c","kernel/signal.c"],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","vendor":"Linux","versions":[{"status":"affected","version":"6.15"},{"lessThan":"6.15","status":"unaffected","version":"0","versionType":"semver"},{"lessThanOrEqual":"6.18.*","status":"unaffected","version":"6.18.54","versionType":"semver"},{"lessThanOrEqual":"7.2.*","status":"unaffected","version":"7.2.8","versionType":"semver"},{"lessThanOrEqual":"*","status":"unaffected","version":"7.3-rc4","versionType":"original_commit_for_fix"}]}],"cpeApplicability":[{"nodes":[{"cpeMatch":[{"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionEndExcluding":"6.18.54","versionStartIncluding":"6.15","vulnerable":true},{"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionEndExcluding":"7.2.8","versionStartIncluding":"6.15","vulnerable":true},{"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionEndExcluding":"7.3-rc4","versionStartIncluding":"6.15","vulnerable":true}],"negate":false,"operator":"OR"}]}],"descriptions":[{"lang":"en","value":"In the Linux kernel, the following vulnerability has been resolved:\n\nsignal: Prevent exec() race\n\nHyunwoo debugged the following KASAN UAF splat:\n\n  BUG: KASAN: slab-use-after-free in __send_signal_locked+0xb27/0xba0\n  Write of size 8 at addr ffff888007ed80c8 by task poc/79\n  ...\n  Call Trace:\n   __send_signal_locked+0xb27/0xba0\n   do_send_sig_info+0xa7/0x160\n   do_send_specific+0x76/0xa0\n   __x64_sys_tgkill+0x193/0x270\n  ...\n  Allocated by task 80:\n   do_timer_create+0x1a4/0x1030\n   __x64_sys_timer_create+0x145/0x190\n  ...\n  Freed by task 12:\n   kmem_cache_free_bulk+0x1f8/0x4a0\n   kvfree_rcu_bulk+0x14f/0x1c0\n   kfree_rcu_work+0x128/0x1a0\n  ...\n  Last potentially related work creation:\n   kvfree_call_rcu+0x39/0x390\n   __flush_itimer_signals+0x211/0x320\n   flush_itimer_signals+0x47/0x90\n   begin_new_exec+0xa6b/0x28c0\n\nIt turned out that this happens with a non-leader exec() as Hyunwoo\nexplained:\n\nde_thread() calls exchange_tids() before release_task(leader), so the\nstruct pid held by a SIGEV_THREAD_ID timer created against the leader's tid\nnow points to the thread which called execve(). pid_task() returns that\nthread and lock_task_sighand() on it succeeds.\n\nIf the timer signal is blocked, its sigqueue stays queued on the leader's\ntask::pending. The next expiry of that timer can then run while\nrelease_task() flushes the queue.\n\nposixtimer_send_sigqueue() checks whether the sigqueue is already queued\nwith a plain list_empty(), which only reads list_head::next.\nlist_del_init() is not atomic and INIT_LIST_HEAD() stores list_head::next\nbefore list_head::prev, so the check can pass in between. list_add_tail()\nqueues the entry on the task::pending of the live thread, and the\nlist_head::prev store from the flush then overwrites the list_head::prev\nlink that list_add_tail() has just set.\n\n__flush_itimer_signals() does not undo that either. With list_head::prev\npointing at the entry itself, its list_del_init() only stores the same\nvalues again, so the entry is not removed from the list. It is still there\nafter the last reference is dropped and the timer is freed by RCU, and the\nlist_add_tail() of a later tgkill() follows that list_head::prev into the\nfreed timer.\n\nThis problem surfaced with the recent commit which moved the sigqueue flush\nout of the sighand lock held region.\n\nHyonwoo proposed to fix this by using list_del_init_careful(), but that\njust papers over the problem. After some disucssions and various attempts\nto solve it, Eric pointed out that there is no reason to flush\ntask::pending late in release_task() and it should be done in\nexit_signals() already.\n\nAs nothing can collect and deliver signals which are queued in a dying\ntask's pending queue, there is no reason to delay it further.\n\nBut it has to be ensured that no signals can be queued into it after that\npoint. exit_signals() sets PF_EXITING in task::flags, which can be used as\nan indicator for this.\n\nCure it by:\n\n  - Preventing signal queueing for task private signals (PIDTYPE_PID) when\n    the task has PF_EXITING set in __send_signal_locked() and in\n    posixtimer_send_sigqueue().\n\n  - Protecting the unlocked setting of PF_EXITING in exit_signals() for the\n    task group empty and the group exit case with sighand lock\n\n  - Flushing task::pending signals right there.\n\n    Optimize that by moving the whole pending list to an on-stack list head\n    under sighand lock and free the signals without the lock held.\n\nThere has been quite some discussion about the lockless flush and the\nnon-leader exec case on weakly ordered systems. The problem is that a third\nparty which tries to send a posix timer signal relies on the PID lookup to\nfind the target task and that lookup might result in the new leader when\nthe signal was originaly directed to the old leader. In case that the\nsignal was queued on the old leader then the lockless flush raised a\nconcern over the following situation:\n\n   old_leader\t\tnew_leader              third party\n\nA: flush_list()\t// list_del_in\n---truncated---"}],"providerMetadata":{"dateUpdated":"2026-10-06T08:45:22.048Z","orgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","shortName":"Linux"},"references":[{"url":"https://git.kernel.org/stable/c/621c365739828900f32a0c1cdfe5387528a62176"},{"url":"https://git.kernel.org/stable/c/934bd95f1a6404dbb845951b823547abab05a649"},{"url":"https://git.kernel.org/stable/c/d2710c8d938ae6a825a6463158e6e6f31eac792a"}],"title":"signal: Prevent exec() race","x_generator":{"engine":"bippy-1.2.0"}}},"cveMetadata":{"assignerOrgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","assignerShortName":"Linux","cveId":"CVE-2026-98256","datePublished":"2026-10-06T08:45:22.048Z","dateReserved":"2026-09-25T10:25:14.331Z","dateUpdated":"2026-10-06T08:45:22.048Z","state":"PUBLISHED"},"dataType":"CVE_RECORD","dataVersion":"5.2"},"nvd":{"publishedDate":"2026-10-06 09:18:14","lastModifiedDate":"2026-10-06 09:18:14","problem_types":[],"metrics":[],"configurations":[]},"legacy_mitre":{"record":{"CveYear":"2026","CveId":"98256","Ordinal":"1","Title":"signal: Prevent exec() race","CVE":"CVE-2026-98256","Year":"2026"},"notes":[{"CveYear":"2026","CveId":"98256","Ordinal":"1","NoteData":"In the Linux kernel, the following vulnerability has been resolved:\n\nsignal: Prevent exec() race\n\nHyunwoo debugged the following KASAN UAF splat:\n\n  BUG: KASAN: slab-use-after-free in __send_signal_locked+0xb27/0xba0\n  Write of size 8 at addr ffff888007ed80c8 by task poc/79\n  ...\n  Call Trace:\n   __send_signal_locked+0xb27/0xba0\n   do_send_sig_info+0xa7/0x160\n   do_send_specific+0x76/0xa0\n   __x64_sys_tgkill+0x193/0x270\n  ...\n  Allocated by task 80:\n   do_timer_create+0x1a4/0x1030\n   __x64_sys_timer_create+0x145/0x190\n  ...\n  Freed by task 12:\n   kmem_cache_free_bulk+0x1f8/0x4a0\n   kvfree_rcu_bulk+0x14f/0x1c0\n   kfree_rcu_work+0x128/0x1a0\n  ...\n  Last potentially related work creation:\n   kvfree_call_rcu+0x39/0x390\n   __flush_itimer_signals+0x211/0x320\n   flush_itimer_signals+0x47/0x90\n   begin_new_exec+0xa6b/0x28c0\n\nIt turned out that this happens with a non-leader exec() as Hyunwoo\nexplained:\n\nde_thread() calls exchange_tids() before release_task(leader), so the\nstruct pid held by a SIGEV_THREAD_ID timer created against the leader's tid\nnow points to the thread which called execve(). pid_task() returns that\nthread and lock_task_sighand() on it succeeds.\n\nIf the timer signal is blocked, its sigqueue stays queued on the leader's\ntask::pending. The next expiry of that timer can then run while\nrelease_task() flushes the queue.\n\nposixtimer_send_sigqueue() checks whether the sigqueue is already queued\nwith a plain list_empty(), which only reads list_head::next.\nlist_del_init() is not atomic and INIT_LIST_HEAD() stores list_head::next\nbefore list_head::prev, so the check can pass in between. list_add_tail()\nqueues the entry on the task::pending of the live thread, and the\nlist_head::prev store from the flush then overwrites the list_head::prev\nlink that list_add_tail() has just set.\n\n__flush_itimer_signals() does not undo that either. With list_head::prev\npointing at the entry itself, its list_del_init() only stores the same\nvalues again, so the entry is not removed from the list. It is still there\nafter the last reference is dropped and the timer is freed by RCU, and the\nlist_add_tail() of a later tgkill() follows that list_head::prev into the\nfreed timer.\n\nThis problem surfaced with the recent commit which moved the sigqueue flush\nout of the sighand lock held region.\n\nHyonwoo proposed to fix this by using list_del_init_careful(), but that\njust papers over the problem. After some disucssions and various attempts\nto solve it, Eric pointed out that there is no reason to flush\ntask::pending late in release_task() and it should be done in\nexit_signals() already.\n\nAs nothing can collect and deliver signals which are queued in a dying\ntask's pending queue, there is no reason to delay it further.\n\nBut it has to be ensured that no signals can be queued into it after that\npoint. exit_signals() sets PF_EXITING in task::flags, which can be used as\nan indicator for this.\n\nCure it by:\n\n  - Preventing signal queueing for task private signals (PIDTYPE_PID) when\n    the task has PF_EXITING set in __send_signal_locked() and in\n    posixtimer_send_sigqueue().\n\n  - Protecting the unlocked setting of PF_EXITING in exit_signals() for the\n    task group empty and the group exit case with sighand lock\n\n  - Flushing task::pending signals right there.\n\n    Optimize that by moving the whole pending list to an on-stack list head\n    under sighand lock and free the signals without the lock held.\n\nThere has been quite some discussion about the lockless flush and the\nnon-leader exec case on weakly ordered systems. The problem is that a third\nparty which tries to send a posix timer signal relies on the PID lookup to\nfind the target task and that lookup might result in the new leader when\nthe signal was originaly directed to the old leader. In case that the\nsignal was queued on the old leader then the lockless flush raised a\nconcern over the following situation:\n\n   old_leader\t\tnew_leader              third party\n\nA: flush_list()\t// list_del_in\n---truncated---","Type":"Description","Title":"signal: Prevent exec() race"}]}}}