[PATCH] sbitmap: Fix sbitmap_put()

[PATCH] sbitmap: Fix sbitmap_put()

[Bart Van Assche]

This patch fixes the following kernel warning:

BUG: using smp_processor_id() in preemptible [00000000] code: scsi_eh_0/152
caller is debug_smp_processor_id+0x17/0x20
CPU: 1 PID: 152 Comm: scsi_eh_0 Tainted: G        W         5.12.0-rc1-dbg+ #6
Call Trace:
 show_stack+0x52/0x58
 dump_stack+0xaf/0xf3
 check_preemption_disabled+0xce/0xd0
 debug_smp_processor_id+0x17/0x20
 scsi_device_unbusy+0x13a/0x1c0 [scsi_mod]
 scsi_finish_command+0x4d/0x290 [scsi_mod]
 scsi_eh_flush_done_q+0x1e7/0x280 [scsi_mod]
 ata_scsi_port_error_handler+0x592/0x750 [libata]
 ata_scsi_error+0x1a0/0x1f0 [libata]
 scsi_error_handler+0x19e/0x330 [scsi_mod]
 kthread+0x222/0x250
 ret_from_fork+0x1f/0x30

Cc: Ming Lei <ming.lei@redhat.com>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Omar Sandoval <osandov@fb.com>
Fixes: c548e62bcf6a ("scsi: sbitmap: Move allocation hint into sbitmap")
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
---
 include/linux/sbitmap.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/linux/sbitmap.h b/include/linux/sbitmap.h
index 3087e1f15fdd..2713e689ad66 100644
--- a/include/linux/sbitmap.h
+++ b/include/linux/sbitmap.h
@@ -324,7 +324,7 @@ static inline void sbitmap_put(struct sbitmap *sb, unsigned int bitnr)
 	sbitmap_deferred_clear_bit(sb, bitnr);
 
 	if (likely(sb->alloc_hint && !sb->round_robin && bitnr < sb->depth))
-		*this_cpu_ptr(sb->alloc_hint) = bitnr;
+		*raw_cpu_ptr(sb->alloc_hint) = bitnr;
 }
 
 static inline int sbitmap_test_bit(struct sbitmap *sb, unsigned int bitnr)

Re: [PATCH] sbitmap: Fix sbitmap_put()

[Ming Lei]

On Mon, Mar 15, 2021 at 08:54:20PM -0700, Bart Van Assche wrote:
> This patch fixes the following kernel warning:
> 
> BUG: using smp_processor_id() in preemptible [00000000] code: scsi_eh_0/152
> caller is debug_smp_processor_id+0x17/0x20
> CPU: 1 PID: 152 Comm: scsi_eh_0 Tainted: G        W         5.12.0-rc1-dbg+ #6
> Call Trace:
>  show_stack+0x52/0x58
>  dump_stack+0xaf/0xf3
>  check_preemption_disabled+0xce/0xd0
>  debug_smp_processor_id+0x17/0x20
>  scsi_device_unbusy+0x13a/0x1c0 [scsi_mod]
>  scsi_finish_command+0x4d/0x290 [scsi_mod]
>  scsi_eh_flush_done_q+0x1e7/0x280 [scsi_mod]
>  ata_scsi_port_error_handler+0x592/0x750 [libata]
>  ata_scsi_error+0x1a0/0x1f0 [libata]
>  scsi_error_handler+0x19e/0x330 [scsi_mod]
>  kthread+0x222/0x250
>  ret_from_fork+0x1f/0x30
> 
> Cc: Ming Lei <ming.lei@redhat.com>
> Cc: Hannes Reinecke <hare@suse.de>
> Cc: Omar Sandoval <osandov@fb.com>
> Fixes: c548e62bcf6a ("scsi: sbitmap: Move allocation hint into sbitmap")
> Signed-off-by: Bart Van Assche <bvanassche@acm.org>
> ---
>  include/linux/sbitmap.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/include/linux/sbitmap.h b/include/linux/sbitmap.h
> index 3087e1f15fdd..2713e689ad66 100644
> --- a/include/linux/sbitmap.h
> +++ b/include/linux/sbitmap.h
> @@ -324,7 +324,7 @@ static inline void sbitmap_put(struct sbitmap *sb, unsigned int bitnr)
>  	sbitmap_deferred_clear_bit(sb, bitnr);
>  
>  	if (likely(sb->alloc_hint && !sb->round_robin && bitnr < sb->depth))
> -		*this_cpu_ptr(sb->alloc_hint) = bitnr;
> +		*raw_cpu_ptr(sb->alloc_hint) = bitnr;
>  }
>  
>  static inline int sbitmap_test_bit(struct sbitmap *sb, unsigned int bitnr)
> 

Reviewed-by: Ming Lei <ming.lei@redhat.com>

-- 
Ming

Re: [PATCH] sbitmap: Fix sbitmap_put()

[Jens Axboe]

On 3/15/21 9:54 PM, Bart Van Assche wrote:
> This patch fixes the following kernel warning:
> 
> BUG: using smp_processor_id() in preemptible [00000000] code: scsi_eh_0/152
> caller is debug_smp_processor_id+0x17/0x20
> CPU: 1 PID: 152 Comm: scsi_eh_0 Tainted: G        W         5.12.0-rc1-dbg+ #6
> Call Trace:
>  show_stack+0x52/0x58
>  dump_stack+0xaf/0xf3
>  check_preemption_disabled+0xce/0xd0
>  debug_smp_processor_id+0x17/0x20
>  scsi_device_unbusy+0x13a/0x1c0 [scsi_mod]
>  scsi_finish_command+0x4d/0x290 [scsi_mod]
>  scsi_eh_flush_done_q+0x1e7/0x280 [scsi_mod]
>  ata_scsi_port_error_handler+0x592/0x750 [libata]
>  ata_scsi_error+0x1a0/0x1f0 [libata]
>  scsi_error_handler+0x19e/0x330 [scsi_mod]
>  kthread+0x222/0x250
>  ret_from_fork+0x1f/0x30

It'd be nice to have a bit of a description in here on what is being
fixed, eg that we're preemptible and cannot safely use this_cpu_ptr(),
and why that it's OK to use raw_cpu_ptr() instead for this particular
application.

-- 
Jens Axboe

Re: [PATCH] sbitmap: Fix sbitmap_put()

[Bart Van Assche]

On 3/16/21 7:32 AM, Jens Axboe wrote:
> It'd be nice to have a bit of a description in here on what is being
> fixed, eg that we're preemptible and cannot safely use this_cpu_ptr(),
> and why that it's OK to use raw_cpu_ptr() instead for this particular
> application.

Sure, I will provide a better description and repost this patch.

Thank you,

Bart.