CVE-2023-53031

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> powerpc/imc-pmu: Fix use of mutex in IRQs disabled section<br /> <br /> Current imc-pmu code triggers a WARNING with CONFIG_DEBUG_ATOMIC_SLEEP<br /> and CONFIG_PROVE_LOCKING enabled, while running a thread_imc event.<br /> <br /> Command to trigger the warning:<br /> # perf stat -e thread_imc/CPM_CS_FROM_L4_MEM_X_DPTEG/ sleep 5<br /> <br /> Performance counter stats for &amp;#39;sleep 5&amp;#39;:<br /> <br /> 0 thread_imc/CPM_CS_FROM_L4_MEM_X_DPTEG/<br /> <br /> 5.002117947 seconds time elapsed<br /> <br /> 0.000131000 seconds user<br /> 0.001063000 seconds sys<br /> <br /> Below is snippet of the warning in dmesg:<br /> <br /> BUG: sleeping function called from invalid context at kernel/locking/mutex.c:580<br /> in_atomic(): 1, irqs_disabled(): 1, non_block: 0, pid: 2869, name: perf-exec<br /> preempt_count: 2, expected: 0<br /> 4 locks held by perf-exec/2869:<br /> #0: c00000004325c540 (&amp;sig-&gt;cred_guard_mutex){+.+.}-{3:3}, at: bprm_execve+0x64/0xa90<br /> #1: c00000004325c5d8 (&amp;sig-&gt;exec_update_lock){++++}-{3:3}, at: begin_new_exec+0x460/0xef0<br /> #2: c0000003fa99d4e0 (&amp;cpuctx_lock){-...}-{2:2}, at: perf_event_exec+0x290/0x510<br /> #3: c000000017ab8418 (&amp;ctx-&gt;lock){....}-{2:2}, at: perf_event_exec+0x29c/0x510<br /> irq event stamp: 4806<br /> hardirqs last enabled at (4805): [] _raw_spin_unlock_irqrestore+0x94/0xd0<br /> hardirqs last disabled at (4806): [] perf_event_exec+0x394/0x510<br /> softirqs last enabled at (0): [] copy_process+0xc34/0x1ff0<br /> softirqs last disabled at (0): [] 0x0<br /> CPU: 36 PID: 2869 Comm: perf-exec Not tainted 6.2.0-rc2-00011-g1247637727f2 #61<br /> Hardware name: 8375-42A POWER9 0x4e1202 opal:v7.0-16-g9b85f7d961 PowerNV<br /> Call Trace:<br /> dump_stack_lvl+0x98/0xe0 (unreliable)<br /> __might_resched+0x2f8/0x310<br /> __mutex_lock+0x6c/0x13f0<br /> thread_imc_event_add+0xf4/0x1b0<br /> event_sched_in+0xe0/0x210<br /> merge_sched_in+0x1f0/0x600<br /> visit_groups_merge.isra.92.constprop.166+0x2bc/0x6c0<br /> ctx_flexible_sched_in+0xcc/0x140<br /> ctx_sched_in+0x20c/0x2a0<br /> ctx_resched+0x104/0x1c0<br /> perf_event_exec+0x340/0x510<br /> begin_new_exec+0x730/0xef0<br /> load_elf_binary+0x3f8/0x1e10<br /> ...<br /> do not call blocking ops when !TASK_RUNNING; state=2001 set at [] do_nanosleep+0x60/0x1a0<br /> WARNING: CPU: 36 PID: 2869 at kernel/sched/core.c:9912 __might_sleep+0x9c/0xb0<br /> CPU: 36 PID: 2869 Comm: sleep Tainted: G W 6.2.0-rc2-00011-g1247637727f2 #61<br /> Hardware name: 8375-42A POWER9 0x4e1202 opal:v7.0-16-g9b85f7d961 PowerNV<br /> NIP: c000000000194a1c LR: c000000000194a18 CTR: c000000000a78670<br /> REGS: c00000004d2134e0 TRAP: 0700 Tainted: G W (6.2.0-rc2-00011-g1247637727f2)<br /> MSR: 9000000000021033 CR: 48002824 XER: 00000000<br /> CFAR: c00000000013fb64 IRQMASK: 1<br /> <br /> The above warning triggered because the current imc-pmu code uses mutex<br /> lock in interrupt disabled sections. The function mutex_lock()<br /> internally calls __might_resched(), which will check if IRQs are<br /> disabled and in case IRQs are disabled, it will trigger the warning.<br /> <br /> Fix the issue by changing the mutex lock to spinlock.<br /> <br /> [mpe: Fix comments, trim oops in change log, add reported-by tags]