| Message ID | 20210619001332.2c0c9a05@xhacker (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | riscv: stacktrace: Fix NULL pointer dereference | expand |
在 2021/6/19 0:13, Jisheng Zhang 写道: > From: Jisheng Zhang <jszhang@kernel.org> > > When CONFIG_FRAME_POINTER=y, calling dump_stack() can always trigger > NULL pointer dereference panic similar as below: > > [ 0.396060] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.13.0-rc5+ #47 > [ 0.396692] Hardware name: riscv-virtio,qemu (DT) > [ 0.397176] Call Trace: > [ 0.398191] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000960 > [ 0.399487] Oops [#1] > [ 0.399739] Modules linked in: > [ 0.400135] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.13.0-rc5+ #47 > [ 0.400570] Hardware name: riscv-virtio,qemu (DT) > [ 0.400926] epc : walk_stackframe+0xc4/0xdc > [ 0.401291] ra : dump_backtrace+0x30/0x38 > [ 0.401630] epc : ffffffff80004922 ra : ffffffff8000496a sp : ffffffe000f3bd00 > [ 0.402115] gp : ffffffff80cfdcb8 tp : ffffffe000f30000 t0 : ffffffff80d0b0cf > [ 0.402602] t1 : ffffffff80d0b0c0 t2 : 0000000000000000 s0 : ffffffe000f3bd60 > [ 0.403071] s1 : ffffffff808bc2e8 a0 : 0000000000001000 a1 : 0000000000000000 > [ 0.403448] a2 : ffffffff803d7088 a3 : ffffffff808bc2e8 a4 : 6131725dbc24d400 > [ 0.403820] a5 : 0000000000001000 a6 : 0000000000000002 a7 : ffffffffffffffff > [ 0.404226] s2 : 0000000000000000 s3 : 0000000000000000 s4 : 0000000000000000 > [ 0.404634] s5 : ffffffff803d7088 s6 : ffffffff808bc2e8 s7 : ffffffff80630650 > [ 0.405085] s8 : ffffffff80912a80 s9 : 0000000000000008 s10: ffffffff804000fc > [ 0.405388] s11: 0000000000000000 t3 : 0000000000000043 t4 : ffffffffffffffff > [ 0.405616] t5 : 000000000000003d t6 : ffffffe000f3baa8 > [ 0.405793] status: 0000000000000100 badaddr: 0000000000000960 cause: 000000000000000d > [ 0.406135] [<ffffffff80004922>] walk_stackframe+0xc4/0xdc > [ 0.407032] [<ffffffff8000496a>] dump_backtrace+0x30/0x38 > [ 0.407797] [<ffffffff803d7100>] show_stack+0x40/0x4c > [ 0.408234] [<ffffffff803d9e5c>] dump_stack+0x90/0xb6 > [ 0.409019] [<ffffffff8040423e>] ptdump_init+0x20/0xc4 > [ 0.409681] [<ffffffff800015b6>] do_one_initcall+0x4c/0x226 > [ 0.410110] [<ffffffff80401094>] kernel_init_freeable+0x1f4/0x258 > [ 0.410562] [<ffffffff803dba88>] kernel_init+0x22/0x148 > [ 0.410959] [<ffffffff800029e2>] ret_from_exception+0x0/0x14 > [ 0.412241] ---[ end trace b2ab92c901b96251 ]--- > [ 0.413099] Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b > > The reason is the task is NULL when we finally call walk_stackframe() > the NULL is passed from __dump_stack(): > > |static void __dump_stack(void) > |{ > | dump_stack_print_info(KERN_DEFAULT); > | show_stack(NULL, NULL, KERN_DEFAULT); > |} > > Fix this issue by checking "task == NULL" case in walk_stackframe(). > > Fixes: eac2f3059e02 ("riscv: stacktrace: fix the riscv stacktrace when CONFIG_FRAME_POINTER enabled" > Signed-off-by: Jisheng Zhang <jszhang@kernel.org> > --- > arch/riscv/kernel/stacktrace.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/arch/riscv/kernel/stacktrace.c b/arch/riscv/kernel/stacktrace.c > index bde85fc53357..7bc8af75933a 100644 > --- a/arch/riscv/kernel/stacktrace.c > +++ b/arch/riscv/kernel/stacktrace.c > @@ -27,7 +27,7 @@ void notrace walk_stackframe(struct task_struct *task, struct pt_regs *regs, > fp = frame_pointer(regs); > sp = user_stack_pointer(regs); > pc = instruction_pointer(regs); > - } else if (task == current) { > + } else if (task == NULL || task == current) { > fp = (unsigned long)__builtin_frame_address(1); > sp = (unsigned long)__builtin_frame_address(0); > pc = (unsigned long)__builtin_return_address(0); > I'm sorry I made a mistake for confusing the function pr_cont's stack with the stack when task is NULL. Anyway, I tested the patch and it looks good. Reviewed-by: Chen Huang <chenhuang5@huawei.com> Thanks.
Hi Palmer, On Sat, 19 Jun 2021 00:13:32 +0800 Jisheng Zhang wrote: > From: Jisheng Zhang <jszhang@kernel.org> > > When CONFIG_FRAME_POINTER=y, calling dump_stack() can always trigger > NULL pointer dereference panic similar as below: > > [ 0.396060] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.13.0-rc5+ #47 > [ 0.396692] Hardware name: riscv-virtio,qemu (DT) > [ 0.397176] Call Trace: > [ 0.398191] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000960 > [ 0.399487] Oops [#1] > [ 0.399739] Modules linked in: > [ 0.400135] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.13.0-rc5+ #47 > [ 0.400570] Hardware name: riscv-virtio,qemu (DT) > [ 0.400926] epc : walk_stackframe+0xc4/0xdc > [ 0.401291] ra : dump_backtrace+0x30/0x38 > [ 0.401630] epc : ffffffff80004922 ra : ffffffff8000496a sp : ffffffe000f3bd00 > [ 0.402115] gp : ffffffff80cfdcb8 tp : ffffffe000f30000 t0 : ffffffff80d0b0cf > [ 0.402602] t1 : ffffffff80d0b0c0 t2 : 0000000000000000 s0 : ffffffe000f3bd60 > [ 0.403071] s1 : ffffffff808bc2e8 a0 : 0000000000001000 a1 : 0000000000000000 > [ 0.403448] a2 : ffffffff803d7088 a3 : ffffffff808bc2e8 a4 : 6131725dbc24d400 > [ 0.403820] a5 : 0000000000001000 a6 : 0000000000000002 a7 : ffffffffffffffff > [ 0.404226] s2 : 0000000000000000 s3 : 0000000000000000 s4 : 0000000000000000 > [ 0.404634] s5 : ffffffff803d7088 s6 : ffffffff808bc2e8 s7 : ffffffff80630650 > [ 0.405085] s8 : ffffffff80912a80 s9 : 0000000000000008 s10: ffffffff804000fc > [ 0.405388] s11: 0000000000000000 t3 : 0000000000000043 t4 : ffffffffffffffff > [ 0.405616] t5 : 000000000000003d t6 : ffffffe000f3baa8 > [ 0.405793] status: 0000000000000100 badaddr: 0000000000000960 cause: 000000000000000d > [ 0.406135] [<ffffffff80004922>] walk_stackframe+0xc4/0xdc > [ 0.407032] [<ffffffff8000496a>] dump_backtrace+0x30/0x38 > [ 0.407797] [<ffffffff803d7100>] show_stack+0x40/0x4c > [ 0.408234] [<ffffffff803d9e5c>] dump_stack+0x90/0xb6 > [ 0.409019] [<ffffffff8040423e>] ptdump_init+0x20/0xc4 > [ 0.409681] [<ffffffff800015b6>] do_one_initcall+0x4c/0x226 > [ 0.410110] [<ffffffff80401094>] kernel_init_freeable+0x1f4/0x258 > [ 0.410562] [<ffffffff803dba88>] kernel_init+0x22/0x148 > [ 0.410959] [<ffffffff800029e2>] ret_from_exception+0x0/0x14 > [ 0.412241] ---[ end trace b2ab92c901b96251 ]--- > [ 0.413099] Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b > > The reason is the task is NULL when we finally call walk_stackframe() > the NULL is passed from __dump_stack(): > > |static void __dump_stack(void) > |{ > | dump_stack_print_info(KERN_DEFAULT); > | show_stack(NULL, NULL, KERN_DEFAULT); > |} > > Fix this issue by checking "task == NULL" case in walk_stackframe(). > > Fixes: eac2f3059e02 ("riscv: stacktrace: fix the riscv stacktrace when CONFIG_FRAME_POINTER enabled" > Signed-off-by: Jisheng Zhang <jszhang@kernel.org> What about this patch? This is a fix to an obvious bug: call dump_stack() will panic. Thanks > --- > arch/riscv/kernel/stacktrace.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/arch/riscv/kernel/stacktrace.c b/arch/riscv/kernel/stacktrace.c > index bde85fc53357..7bc8af75933a 100644 > --- a/arch/riscv/kernel/stacktrace.c > +++ b/arch/riscv/kernel/stacktrace.c > @@ -27,7 +27,7 @@ void notrace walk_stackframe(struct task_struct *task, struct pt_regs *regs, > fp = frame_pointer(regs); > sp = user_stack_pointer(regs); > pc = instruction_pointer(regs); > - } else if (task == current) { > + } else if (task == NULL || task == current) { > fp = (unsigned long)__builtin_frame_address(1); > sp = (unsigned long)__builtin_frame_address(0); > pc = (unsigned long)__builtin_return_address(0);
diff --git a/arch/riscv/kernel/stacktrace.c b/arch/riscv/kernel/stacktrace.c index bde85fc53357..7bc8af75933a 100644 --- a/arch/riscv/kernel/stacktrace.c +++ b/arch/riscv/kernel/stacktrace.c @@ -27,7 +27,7 @@ void notrace walk_stackframe(struct task_struct *task, struct pt_regs *regs, fp = frame_pointer(regs); sp = user_stack_pointer(regs); pc = instruction_pointer(regs); - } else if (task == current) { + } else if (task == NULL || task == current) { fp = (unsigned long)__builtin_frame_address(1); sp = (unsigned long)__builtin_frame_address(0); pc = (unsigned long)__builtin_return_address(0);