Skip to contentSkip to conversation

riscv: stacktrace: fix dump_backtrace/walk_stackframe with NULL task

PATCH In review Changbin Du 3 participants 5 messages Jun 27, 2021 – Jul 28, 2021 RISC-V

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 riscv: stacktrace: fix dump_backtrace/walk_stackframe with NULL task

Some places try to show backtrace with NULL task, and expect the task is 'current'. For example, dump_stack()->show_stack(NULL,...). So the stacktrace code should take care of this case.

Here is an oops caused by this issue when accessing the NULL task.

[   15.180813] Kernel panic - not syncing: No working init found.  Try passing init= option to kernel. See Linux Documentation/admin-guide/init.rst for guidance.
[   15.182382] CPU: 3 PID: 1 Comm: swapper/0 Not tainted 5.13.0-rc7-00111-g625acffd7ae2-dirty #18
[   15.183431] Hardware name: riscv-virtio,qemu (DT)
[   15.184253] Call Trace:
[   15.223617] Unable to handle kernel paging request at virtual address 0000000000001590
[   15.267378] Oops [#1]
[   15.268215] Modules linked in:
[   15.272027] CPU: 3 PID: 1 Comm: swapper/0 Not tainted 5.13.0-rc7-00111-g625acffd7ae2-dirty #18
[   15.273997] Hardware name: riscv-virtio,qemu (DT)
[   15.275134] epc : walk_stackframe+0xc4/0xdc
[   15.280146]  ra : dump_backtrace+0x30/0x38
[   15.280799] epc : ffffffff8000597e ra : ffffffff800059c6 sp : ffffffe002383d60
[   15.281622]  gp : ffffffff8179ad18 tp : ffffffe002378000 t0 : ffffffff81bc1a3f
[   15.282574]  t1 : 0000000000000001 t2 : 0000000000000000 s0 : ffffffe002383dc0
[   15.283782]  s1 : ffffffff812b7d18 a0 : 0000000000001000 a1 : 0000000000000000
[   15.285115]  a2 : ffffffff807ec668 a3 : ffffffff812b7d18 a4 : c76c00cabf08b500
[   15.286213]  a5 : 0000000000001000 a6 : 000000001a9ef260 a7 : 0000000000000000
[   15.287317]  s2 : 0000000000000000 s3 : 0000000000000000 s4 : 0000000000000000
[   15.288323]  s5 : ffffffff807ec668 s6 : ffffffff812b7d18 s7 : 0000000000000000
[   15.289530]  s8 : 0000000000000000 s9 : 0000000000000000 s10: 0000000000000000
[   15.290995]  s11: 0000000000000000 t3 : 0000000000000001 t4 : 0000000000000000
[   15.292465]  t5 : 206f74206e6f6974 t6 : ffffffe002383b28
[   15.293859] status: 0000000000000100 badaddr: 0000000000001590 cause: 000000000000000d
[   15.296035] [<ffffffff8000597e>] walk_stackframe+0xc4/0xdc
[   15.297342] [<ffffffff800059c6>] dump_backtrace+0x30/0x38
[   15.298333] [<ffffffff807ec6e0>] show_stack+0x40/0x4c
[   15.299765] [<ffffffff807f07ac>] dump_stack+0x7c/0x96
[   15.300553] [<ffffffff807ec8be>] panic+0x118/0x300
[   15.301147] [<ffffffff807f61e8>] kernel_init+0x12c/0x138
[   15.302056] [<ffffffff80003a22>] ret_from_exception+0x0/0xc
[   15.338628] ---[ end trace 0a3fa0cc7f3393cd ]---
[   15.339919] note: swapper/0[1] exited with preempt_count 1
[   15.341995] Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b
[   15.343889] SMP: stopping secondary CPUs
[   16.802836] SMP: failed to stop secondary CPUs 0-3
[   16.806264] ---[ end Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b ]---

This patch fixes it by setting the task to current if it's NULL before accessing it.

Fixes 5d8544e2d0 ("RISC-V: Generic library routines and assembly")
---
 arch/riscv/kernel/stacktrace.c | 6 ++++++
 1 file changed, 6 insertions(+)
arch/riscv/kernel/stacktrace.c +6 -0
index bde85fc53357..788b65eba965 100644
@@ -23,6 +23,9 @@ void notrace walk_stackframe(struct task_struct *task, struct pt_regs *regs,
 {
 	unsigned long fp, sp, pc;
 
+	if (!task)
+		task = current;
+
 	if (regs) {
 		fp = frame_pointer(regs);
 		sp = user_stack_pointer(regs);
@@ -73,6 +76,9 @@ void notrace walk_stackframe(struct task_struct *task,
 	unsigned long sp, pc;
 	unsigned long *ksp;
 
+	if (!task)
+		task = current;
+
 	if (regs) {
 		sp = user_stack_pointer(regs);
 		pc = instruction_pointer(regs);
Signature · 1 line
2.30.2
#1 lore raw
Quoted 5 lines · On Sun, 27 Jun 2021 17:26:59 +0800 Changbin Du <[email protected]> wrote:

Some places try to show backtrace with NULL task, and expect the task is 'current'. For example, dump_stack()->show_stack(NULL,...). So the stacktrace code should take care of this case.

I fixed this issue one week ago:

http://lists.infradead.org/pipermail/linux-riscv/2021-June/007258.html

Quoted 40 lines

Here is an oops caused by this issue when accessing the NULL task.

[   15.180813] Kernel panic - not syncing: No working init found.  Try passing init= option to kernel. See Linux Documentation/admin-guide/init.rst for guidance.
[   15.182382] CPU: 3 PID: 1 Comm: swapper/0 Not tainted 5.13.0-rc7-00111-g625acffd7ae2-dirty #18
[   15.183431] Hardware name: riscv-virtio,qemu (DT)
[   15.184253] Call Trace:
[   15.223617] Unable to handle kernel paging request at virtual address 0000000000001590
[   15.267378] Oops [#1]
[   15.268215] Modules linked in:
[   15.272027] CPU: 3 PID: 1 Comm: swapper/0 Not tainted 5.13.0-rc7-00111-g625acffd7ae2-dirty #18
[   15.273997] Hardware name: riscv-virtio,qemu (DT)
[   15.275134] epc : walk_stackframe+0xc4/0xdc
[   15.280146]  ra : dump_backtrace+0x30/0x38
[   15.280799] epc : ffffffff8000597e ra : ffffffff800059c6 sp : ffffffe002383d60
[   15.281622]  gp : ffffffff8179ad18 tp : ffffffe002378000 t0 : ffffffff81bc1a3f
[   15.282574]  t1 : 0000000000000001 t2 : 0000000000000000 s0 : ffffffe002383dc0
[   15.283782]  s1 : ffffffff812b7d18 a0 : 0000000000001000 a1 : 0000000000000000
[   15.285115]  a2 : ffffffff807ec668 a3 : ffffffff812b7d18 a4 : c76c00cabf08b500
[   15.286213]  a5 : 0000000000001000 a6 : 000000001a9ef260 a7 : 0000000000000000
[   15.287317]  s2 : 0000000000000000 s3 : 0000000000000000 s4 : 0000000000000000
[   15.288323]  s5 : ffffffff807ec668 s6 : ffffffff812b7d18 s7 : 0000000000000000
[   15.289530]  s8 : 0000000000000000 s9 : 0000000000000000 s10: 0000000000000000
[   15.290995]  s11: 0000000000000000 t3 : 0000000000000001 t4 : 0000000000000000
[   15.292465]  t5 : 206f74206e6f6974 t6 : ffffffe002383b28
[   15.293859] status: 0000000000000100 badaddr: 0000000000001590 cause: 000000000000000d
[   15.296035] [<ffffffff8000597e>] walk_stackframe+0xc4/0xdc
[   15.297342] [<ffffffff800059c6>] dump_backtrace+0x30/0x38
[   15.298333] [<ffffffff807ec6e0>] show_stack+0x40/0x4c
[   15.299765] [<ffffffff807f07ac>] dump_stack+0x7c/0x96
[   15.300553] [<ffffffff807ec8be>] panic+0x118/0x300
[   15.301147] [<ffffffff807f61e8>] kernel_init+0x12c/0x138
[   15.302056] [<ffffffff80003a22>] ret_from_exception+0x0/0xc
[   15.338628] ---[ end trace 0a3fa0cc7f3393cd ]---
[   15.339919] note: swapper/0[1] exited with preempt_count 1
[   15.341995] Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b
[   15.343889] SMP: stopping secondary CPUs
[   16.802836] SMP: failed to stop secondary CPUs 0-3
[   16.806264] ---[ end Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b ]---
Quoted 5 lines

This patch fixes it by setting the task to current if it's NULL before accessing it.

Fixes 5d8544e2d0 ("RISC-V: Generic library routines and assembly")

Hmm, this fixes tag should be

Fixes eac2f3059e02 ("riscv: stacktrace: fix the riscv stacktrace when CONFIG_FRAME_POINTER enabled"
Quoted 35 lines
---
 arch/riscv/kernel/stacktrace.c | 6 ++++++
 1 file changed, 6 insertions(+)
arch/riscv/kernel/stacktrace.c +6 -1
index bde85fc53357..788b65eba965 100644
@@ -23,6 +23,9 @@ void notrace walk_stackframe(struct task_struct *task, struct pt_regs *regs,
 {
        unsigned long fp, sp, pc;
 
+       if (!task)
+               task = current;
+
        if (regs) {
                fp = frame_pointer(regs);
                sp = user_stack_pointer(regs);
@@ -73,6 +76,9 @@ void notrace walk_stackframe(struct task_struct *task,
        unsigned long sp, pc;
        unsigned long *ksp;
 
+       if (!task)
+               task = current;
+
        if (regs) {
                sp = user_stack_pointer(regs);
                pc = instruction_pointer(regs);
--

2.30.2

_______________________________________________
linux-riscv mailing list
[email protected]

#2 lore raw
Quoted 6 lines · Jisheng Zhang
Quoted 2 lines · On Sun, 27 Jun 2021 17:26:59 +0800 Changbin Du <[email protected]> wrote:
Quoted 8 lines
Quoted 3 lines

Some places try to show backtrace with NULL task, and expect the task is 'current'. For example, dump_stack()->show_stack(NULL,...). So the stacktrace code should take care of this case.

I fixed this issue one week ago:

http://lists.infradead.org/pipermail/linux-riscv/2021-June/007258.html

Good to know. Thanks!

Quoted 86 lines
Quoted 40 lines

Here is an oops caused by this issue when accessing the NULL task.

[   15.180813] Kernel panic - not syncing: No working init found.  Try passing init= option to kernel. See Linux Documentation/admin-guide/init.rst for guidance.
[   15.182382] CPU: 3 PID: 1 Comm: swapper/0 Not tainted 5.13.0-rc7-00111-g625acffd7ae2-dirty #18
[   15.183431] Hardware name: riscv-virtio,qemu (DT)
[   15.184253] Call Trace:
[   15.223617] Unable to handle kernel paging request at virtual address 0000000000001590
[   15.267378] Oops [#1]
[   15.268215] Modules linked in:
[   15.272027] CPU: 3 PID: 1 Comm: swapper/0 Not tainted 5.13.0-rc7-00111-g625acffd7ae2-dirty #18
[   15.273997] Hardware name: riscv-virtio,qemu (DT)
[   15.275134] epc : walk_stackframe+0xc4/0xdc
[   15.280146]  ra : dump_backtrace+0x30/0x38
[   15.280799] epc : ffffffff8000597e ra : ffffffff800059c6 sp : ffffffe002383d60
[   15.281622]  gp : ffffffff8179ad18 tp : ffffffe002378000 t0 : ffffffff81bc1a3f
[   15.282574]  t1 : 0000000000000001 t2 : 0000000000000000 s0 : ffffffe002383dc0
[   15.283782]  s1 : ffffffff812b7d18 a0 : 0000000000001000 a1 : 0000000000000000
[   15.285115]  a2 : ffffffff807ec668 a3 : ffffffff812b7d18 a4 : c76c00cabf08b500
[   15.286213]  a5 : 0000000000001000 a6 : 000000001a9ef260 a7 : 0000000000000000
[   15.287317]  s2 : 0000000000000000 s3 : 0000000000000000 s4 : 0000000000000000
[   15.288323]  s5 : ffffffff807ec668 s6 : ffffffff812b7d18 s7 : 0000000000000000
[   15.289530]  s8 : 0000000000000000 s9 : 0000000000000000 s10: 0000000000000000
[   15.290995]  s11: 0000000000000000 t3 : 0000000000000001 t4 : 0000000000000000
[   15.292465]  t5 : 206f74206e6f6974 t6 : ffffffe002383b28
[   15.293859] status: 0000000000000100 badaddr: 0000000000001590 cause: 000000000000000d
[   15.296035] [<ffffffff8000597e>] walk_stackframe+0xc4/0xdc
[   15.297342] [<ffffffff800059c6>] dump_backtrace+0x30/0x38
[   15.298333] [<ffffffff807ec6e0>] show_stack+0x40/0x4c
[   15.299765] [<ffffffff807f07ac>] dump_stack+0x7c/0x96
[   15.300553] [<ffffffff807ec8be>] panic+0x118/0x300
[   15.301147] [<ffffffff807f61e8>] kernel_init+0x12c/0x138
[   15.302056] [<ffffffff80003a22>] ret_from_exception+0x0/0xc
[   15.338628] ---[ end trace 0a3fa0cc7f3393cd ]---
[   15.339919] note: swapper/0[1] exited with preempt_count 1
[   15.341995] Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b
[   15.343889] SMP: stopping secondary CPUs
[   16.802836] SMP: failed to stop secondary CPUs 0-3
[   16.806264] ---[ end Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b ]---
Quoted 5 lines

This patch fixes it by setting the task to current if it's NULL before accessing it.

Fixes 5d8544e2d0 ("RISC-V: Generic library routines and assembly")

Hmm, this fixes tag should be

Fixes eac2f3059e02 ("riscv: stacktrace: fix the riscv stacktrace when CONFIG_FRAME_POINTER enabled"
Quoted 35 lines
---
 arch/riscv/kernel/stacktrace.c | 6 ++++++
 1 file changed, 6 insertions(+)
arch/riscv/kernel/stacktrace.c +6 -1
index bde85fc53357..788b65eba965 100644
@@ -23,6 +23,9 @@ void notrace walk_stackframe(struct task_struct *task, struct pt_regs *regs,
 {
        unsigned long fp, sp, pc;
 
+       if (!task)
+               task = current;
+
        if (regs) {
                fp = frame_pointer(regs);
                sp = user_stack_pointer(regs);
@@ -73,6 +76,9 @@ void notrace walk_stackframe(struct task_struct *task,
        unsigned long sp, pc;
        unsigned long *ksp;
 
+       if (!task)
+               task = current;
+
        if (regs) {
                sp = user_stack_pointer(regs);
                pc = instruction_pointer(regs);
--

2.30.2

_______________________________________________
linux-riscv mailing list
[email protected]

Signature · 2 lines
Cheers,
Changbin Du
#3 lore raw
In reply to Jisheng Zhang
Quoted 6 lines · Jisheng Zhang
Quoted 2 lines · On Sun, 27 Jun 2021 17:26:59 +0800 Changbin Du <[email protected]> wrote:
Quoted 7 lines
Quoted 3 lines

Some places try to show backtrace with NULL task, and expect the task is 'current'. For example, dump_stack()->show_stack(NULL,...). So the stacktrace code should take care of this case.

I fixed this issue one week ago:

http://lists.infradead.org/pipermail/linux-riscv/2021-June/007258.html

I still see this issue on mainline. Is your fix merged? Thanks!

Signature · 2 lines
Cheers,
Changbin Du
#4 lore raw
Quoted 11 lines · On Wed, 28 Jul 2021 06:16:56 +0800 Changbin Du <[email protected]> wrote:
Quoted 10 lines · Jisheng Zhang
Quoted 5 lines · On Sun, 27 Jun 2021 17:26:59 +0800 Changbin Du <[email protected]> wrote:

Some places try to show backtrace with NULL task, and expect the task is 'current'. For example, dump_stack()->show_stack(NULL,...). So the stacktrace code should take care of this case.

Quoted 5 lines
Quoted 3 lines

I fixed this issue one week ago:

http://lists.infradead.org/pipermail/linux-riscv/2021-June/007258.html

I still see this issue on mainline. Is your fix merged? Thanks!

Nope, the fix is missed twice. Palmer has added the fix patch into fix branch, I help it will be in next rc

Regards

_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv