diff mbox series

samples/hw_breakpoint: Fix kernel BUG 'invalid opcode: 0000'

Message ID tencent_E626A858BED28C4E21C219780BC566015D0A@qq.com
State New
Headers
Series samples/hw_breakpoint: Fix kernel BUG 'invalid opcode: 0000' |

Commit Message

Rong Tao July 24, 2023, 10:47 a.m. UTC 
  From: Rong Tao <rongtao@cestc.cn>

Macro symbol_put() is defined as __symbol_put(__stringify(x))

    ksym_name = "jiffies"
    symbol_put(ksym_name)

will be resolved as

    __symbol_put("ksym_name")

which is clearly wrong. So symbol_put must be replaced with __symbol_put.

When we uninstall hw_breakpoint.ko (rmmod), a kernel bug occurs with the
following error:

[11381.854152] kernel BUG at kernel/module/main.c:779!
[11381.854159] invalid opcode: 0000 [#2] PREEMPT SMP PTI
[11381.854163] CPU: 8 PID: 59623 Comm: rmmod Tainted: G      D    OE      6.2.9-200.fc37.x86_64 #1
[11381.854167] Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./B360M-HDV, BIOS P3.20 10/23/2018
[11381.854169] RIP: 0010:__symbol_put+0xa2/0xb0
[11381.854175] Code: 00 e8 92 d2 f7 ff 65 8b 05 c3 2f e6 78 85 c0 74 1b 48 8b 44 24 30 65 48 2b 04 25 28 00 00 00 75 12 48 83 c4 38 c3 cc cc cc cc <0f> 0b 0f 1f 44 00 00 eb de e8 c0 df d8 00 90 90 90 90 90 90 90 90
[11381.854178] RSP: 0018:ffffad8ec6ae7dd0 EFLAGS: 00010246
[11381.854181] RAX: 0000000000000000 RBX: ffffffffc1fd1240 RCX: 000000000000000c
[11381.854184] RDX: 000000000000006b RSI: ffffffffc02bf7c7 RDI: ffffffffc1fd001c
[11381.854186] RBP: 000055a38b76e7c8 R08: ffffffff871ccfe0 R09: 0000000000000000
[11381.854188] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
[11381.854190] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
[11381.854192] FS:  00007fbf7c62c740(0000) GS:ffff8c5badc00000(0000) knlGS:0000000000000000
[11381.854195] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[11381.854197] CR2: 000055a38b7793f8 CR3: 0000000363e1e001 CR4: 00000000003726e0
[11381.854200] DR0: ffffffffb3407980 DR1: 0000000000000000 DR2: 0000000000000000
[11381.854202] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
[11381.854204] Call Trace:
[11381.854207]  <TASK>
[11381.854212]  s_module_exit+0xc/0xff0 [symbol_getput]
[11381.854219]  __do_sys_delete_module.constprop.0+0x198/0x2f0
[11381.854225]  do_syscall_64+0x58/0x80
[11381.854231]  ? exit_to_user_mode_prepare+0x180/0x1f0
[11381.854237]  ? syscall_exit_to_user_mode+0x17/0x40
[11381.854241]  ? do_syscall_64+0x67/0x80
[11381.854245]  ? syscall_exit_to_user_mode+0x17/0x40
[11381.854248]  ? do_syscall_64+0x67/0x80
[11381.854252]  ? exc_page_fault+0x70/0x170
[11381.854256]  entry_SYSCALL_64_after_hwframe+0x72/0xdc

Signed-off-by: Rong Tao <rongtao@cestc.cn>
---
 samples/hw_breakpoint/data_breakpoint.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Petr Mladek July 24, 2023, 12:59 p.m. UTC | #1 
On Mon 2023-07-24 18:47:58, Rong Tao wrote:
> From: Rong Tao <rongtao@cestc.cn>
> 
> Macro symbol_put() is defined as __symbol_put(__stringify(x))
> 
>     ksym_name = "jiffies"
>     symbol_put(ksym_name)
> 
> will be resolved as
> 
>     __symbol_put("ksym_name")
> 
> which is clearly wrong. So symbol_put must be replaced with __symbol_put.
> 
> When we uninstall hw_breakpoint.ko (rmmod), a kernel bug occurs with the
> following error:
> 
> [11381.854152] kernel BUG at kernel/module/main.c:779!
> [11381.854159] invalid opcode: 0000 [#2] PREEMPT SMP PTI
> [11381.854163] CPU: 8 PID: 59623 Comm: rmmod Tainted: G      D    OE      6.2.9-200.fc37.x86_64 #1
> [11381.854167] Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./B360M-HDV, BIOS P3.20 10/23/2018
> [11381.854169] RIP: 0010:__symbol_put+0xa2/0xb0
> [11381.854175] Code: 00 e8 92 d2 f7 ff 65 8b 05 c3 2f e6 78 85 c0 74 1b 48 8b 44 24 30 65 48 2b 04 25 28 00 00 00 75 12 48 83 c4 38 c3 cc cc cc cc <0f> 0b 0f 1f 44 00 00 eb de e8 c0 df d8 00 90 90 90 90 90 90 90 90
> [11381.854178] RSP: 0018:ffffad8ec6ae7dd0 EFLAGS: 00010246
> [11381.854181] RAX: 0000000000000000 RBX: ffffffffc1fd1240 RCX: 000000000000000c
> [11381.854184] RDX: 000000000000006b RSI: ffffffffc02bf7c7 RDI: ffffffffc1fd001c
> [11381.854186] RBP: 000055a38b76e7c8 R08: ffffffff871ccfe0 R09: 0000000000000000
> [11381.854188] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
> [11381.854190] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
> [11381.854192] FS:  00007fbf7c62c740(0000) GS:ffff8c5badc00000(0000) knlGS:0000000000000000
> [11381.854195] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [11381.854197] CR2: 000055a38b7793f8 CR3: 0000000363e1e001 CR4: 00000000003726e0
> [11381.854200] DR0: ffffffffb3407980 DR1: 0000000000000000 DR2: 0000000000000000
> [11381.854202] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
> [11381.854204] Call Trace:
> [11381.854207]  <TASK>
> [11381.854212]  s_module_exit+0xc/0xff0 [symbol_getput]
> [11381.854219]  __do_sys_delete_module.constprop.0+0x198/0x2f0
> [11381.854225]  do_syscall_64+0x58/0x80
> [11381.854231]  ? exit_to_user_mode_prepare+0x180/0x1f0
> [11381.854237]  ? syscall_exit_to_user_mode+0x17/0x40
> [11381.854241]  ? do_syscall_64+0x67/0x80
> [11381.854245]  ? syscall_exit_to_user_mode+0x17/0x40
> [11381.854248]  ? do_syscall_64+0x67/0x80
> [11381.854252]  ? exc_page_fault+0x70/0x170
> [11381.854256]  entry_SYSCALL_64_after_hwframe+0x72/0xdc
> 
> Signed-off-by: Rong Tao <rongtao@cestc.cn>

Reviewed-by: Petr Mladek <pmladek@suse.com>

I have already seen and acked this patch few months ago, see
https://lore.kernel.org/all/ZD0TfQHWQftNvFNA@alley/#t

symbol_put() is in module loader code, so this might go via
the module loaded tree. Adding Luis into Cc.

Best Regards,
Petr

---
>  samples/hw_breakpoint/data_breakpoint.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/samples/hw_breakpoint/data_breakpoint.c b/samples/hw_breakpoint/data_breakpoint.c
> index 418c46fe5ffc..9debd128b2ab 100644
> --- a/samples/hw_breakpoint/data_breakpoint.c
> +++ b/samples/hw_breakpoint/data_breakpoint.c
> @@ -70,7 +70,7 @@ static int __init hw_break_module_init(void)
>  static void __exit hw_break_module_exit(void)
>  {
>  	unregister_wide_hw_breakpoint(sample_hbp);
> -	symbol_put(ksym_name);
> +	__symbol_put(ksym_name);
>  	printk(KERN_INFO "HW Breakpoint for %s write uninstalled\n", ksym_name);
>  }
>  
> -- 
> 2.39.3
Luis Chamberlain July 24, 2023, 7:16 p.m. UTC | #2 
On Mon, Jul 24, 2023 at 02:59:15PM +0200, Petr Mladek wrote:
> On Mon 2023-07-24 18:47:58, Rong Tao wrote:
> > From: Rong Tao <rongtao@cestc.cn>
> > 
> > Macro symbol_put() is defined as __symbol_put(__stringify(x))
> > 
> >     ksym_name = "jiffies"
> >     symbol_put(ksym_name)
> > 
> > will be resolved as
> > 
> >     __symbol_put("ksym_name")
> > 
> > which is clearly wrong. So symbol_put must be replaced with __symbol_put.
> > 
> > When we uninstall hw_breakpoint.ko (rmmod), a kernel bug occurs with the
> > following error:
> > 
> > [11381.854152] kernel BUG at kernel/module/main.c:779!
> > [11381.854159] invalid opcode: 0000 [#2] PREEMPT SMP PTI
> > [11381.854163] CPU: 8 PID: 59623 Comm: rmmod Tainted: G      D    OE      6.2.9-200.fc37.x86_64 #1
> > [11381.854167] Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./B360M-HDV, BIOS P3.20 10/23/2018
> > [11381.854169] RIP: 0010:__symbol_put+0xa2/0xb0
> > [11381.854175] Code: 00 e8 92 d2 f7 ff 65 8b 05 c3 2f e6 78 85 c0 74 1b 48 8b 44 24 30 65 48 2b 04 25 28 00 00 00 75 12 48 83 c4 38 c3 cc cc cc cc <0f> 0b 0f 1f 44 00 00 eb de e8 c0 df d8 00 90 90 90 90 90 90 90 90
> > [11381.854178] RSP: 0018:ffffad8ec6ae7dd0 EFLAGS: 00010246
> > [11381.854181] RAX: 0000000000000000 RBX: ffffffffc1fd1240 RCX: 000000000000000c
> > [11381.854184] RDX: 000000000000006b RSI: ffffffffc02bf7c7 RDI: ffffffffc1fd001c
> > [11381.854186] RBP: 000055a38b76e7c8 R08: ffffffff871ccfe0 R09: 0000000000000000
> > [11381.854188] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
> > [11381.854190] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
> > [11381.854192] FS:  00007fbf7c62c740(0000) GS:ffff8c5badc00000(0000) knlGS:0000000000000000
> > [11381.854195] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> > [11381.854197] CR2: 000055a38b7793f8 CR3: 0000000363e1e001 CR4: 00000000003726e0
> > [11381.854200] DR0: ffffffffb3407980 DR1: 0000000000000000 DR2: 0000000000000000
> > [11381.854202] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
> > [11381.854204] Call Trace:
> > [11381.854207]  <TASK>
> > [11381.854212]  s_module_exit+0xc/0xff0 [symbol_getput]
> > [11381.854219]  __do_sys_delete_module.constprop.0+0x198/0x2f0
> > [11381.854225]  do_syscall_64+0x58/0x80
> > [11381.854231]  ? exit_to_user_mode_prepare+0x180/0x1f0
> > [11381.854237]  ? syscall_exit_to_user_mode+0x17/0x40
> > [11381.854241]  ? do_syscall_64+0x67/0x80
> > [11381.854245]  ? syscall_exit_to_user_mode+0x17/0x40
> > [11381.854248]  ? do_syscall_64+0x67/0x80
> > [11381.854252]  ? exc_page_fault+0x70/0x170
> > [11381.854256]  entry_SYSCALL_64_after_hwframe+0x72/0xdc
> > 
> > Signed-off-by: Rong Tao <rongtao@cestc.cn>
> 
> Reviewed-by: Petr Mladek <pmladek@suse.com>

Applied and pushed, as this is just samples/ directory I won't send
right away to Linus as its not that critical.

  Luis
Rong Tao July 25, 2023, 1:38 p.m. UTC | #3 
Hi Luis

The 'kernel test robot' report an error [0]:

	>> samples/hw_breakpoint/data_breakpoint.c:73:2: error: call to undeclared function '__symbol_put'; 
	ISO C99 and later do not support implicit function declarations [-Werror,-Wimplicit-function-declaration]
	           __symbol_put(ksym_name);
	           ^
	   samples/hw_breakpoint/data_breakpoint.c:73:2: note: did you mean '__symbol_get'?
	   include/linux/module.h:301:7: note: '__symbol_get' declared here
	   void *__symbol_get(const char *symbol);
	         ^
	   1 error generated.

Do I need to do something? It seems that no changes need to be made from the
mainline code.

[0] https://lore.kernel.org/oe-kbuild-all/202307250926.06DQZjVe-lkp@intel.com/

Rong
diff mbox series

Patch

diff --git a/samples/hw_breakpoint/data_breakpoint.c b/samples/hw_breakpoint/data_breakpoint.c
index 418c46fe5ffc..9debd128b2ab 100644
--- a/samples/hw_breakpoint/data_breakpoint.c
+++ b/samples/hw_breakpoint/data_breakpoint.c
@@ -70,7 +70,7 @@  static int __init hw_break_module_init(void)
 static void __exit hw_break_module_exit(void)
 {
 	unregister_wide_hw_breakpoint(sample_hbp);
-	symbol_put(ksym_name);
+	__symbol_put(ksym_name);
 	printk(KERN_INFO "HW Breakpoint for %s write uninstalled\n", ksym_name);
 }