[PATCH] block: Add should_fail_bio() for bpf error injection

[PATCH] block: Add should_fail_bio() for bpf error injection

[Howard McLauchlan]

The classic error injection mechanism, should_fail_request() does not
support use cases where more information is required (from the entire
struct bio, for example).

To that end, this patch introduces should_fail_bio(), which calls
should_fail_request() under the hood but provides a convenient
place for kprobes to hook into if they require the entire struct bio.
This patch also replaces some existing calls to should_fail_request()
with should_fail_bio() with no degradation in performance.

Signed-off-by: Howard McLauchlan <hmclauchlan@fb.com>
---
 block/blk-core.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/block/blk-core.c b/block/blk-core.c
index b8881750a3ac..5e73c996d338 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -34,6 +34,7 @@
 #include <linux/pm_runtime.h>
 #include <linux/blk-cgroup.h>
 #include <linux/debugfs.h>
+#include <linux/bpf.h>
 
 #define CREATE_TRACE_POINTS
 #include <trace/events/block.h>
@@ -2050,6 +2051,12 @@ static inline bool should_fail_request(struct hd_struct *part,
 
 #endif /* CONFIG_FAIL_MAKE_REQUEST */
 
+static noinline bool should_fail_bio(struct bio *bio)
+{
+	return should_fail_request(&bio->bi_disk->part0, bio->bi_iter.bi_size);
+}
+BPF_ALLOW_ERROR_INJECTION(should_fail_bio);
+
 /*
  * Remap block n of partition p to block n+start(p) of the disk.
  */
@@ -2141,7 +2148,7 @@ generic_make_request_checks(struct bio *bio)
 	if ((bio->bi_opf & REQ_NOWAIT) && !queue_is_rq_based(q))
 		goto not_supported;
 
-	if (should_fail_request(&bio->bi_disk->part0, bio->bi_iter.bi_size))
+	if (should_fail_bio(bio))
 		goto end_io;
 
 	if (blk_partition_remap(bio))
-- 
2.14.1

Re: [PATCH] block: Add should_fail_bio() for bpf error injection

[Omar Sandoval]

On Wed, Jan 24, 2018 at 03:22:58PM -0800, Howard McLauchlan wrote:
> The classic error injection mechanism, should_fail_request() does not
> support use cases where more information is required (from the entire
> struct bio, for example).
> 
> To that end, this patch introduces should_fail_bio(), which calls
> should_fail_request() under the hood but provides a convenient
> place for kprobes to hook into if they require the entire struct bio.
> This patch also replaces some existing calls to should_fail_request()
> with should_fail_bio() with no degradation in performance.

Reviewed-by: Omar Sandoval <osandov@fb.com>

Jens, can we pick this up for 4.16? The necessary BPF support has been
merged into Linus' tree.

> Signed-off-by: Howard McLauchlan <hmclauchlan@fb.com>
> ---
>  block/blk-core.c | 9 ++++++++-
>  1 file changed, 8 insertions(+), 1 deletion(-)
> 
> diff --git a/block/blk-core.c b/block/blk-core.c
> index b8881750a3ac..5e73c996d338 100644
> --- a/block/blk-core.c
> +++ b/block/blk-core.c
> @@ -34,6 +34,7 @@
>  #include <linux/pm_runtime.h>
>  #include <linux/blk-cgroup.h>
>  #include <linux/debugfs.h>
> +#include <linux/bpf.h>
>  
>  #define CREATE_TRACE_POINTS
>  #include <trace/events/block.h>
> @@ -2050,6 +2051,12 @@ static inline bool should_fail_request(struct hd_struct *part,
>  
>  #endif /* CONFIG_FAIL_MAKE_REQUEST */
>  
> +static noinline bool should_fail_bio(struct bio *bio)
> +{
> +	return should_fail_request(&bio->bi_disk->part0, bio->bi_iter.bi_size);
> +}
> +BPF_ALLOW_ERROR_INJECTION(should_fail_bio);
> +
>  /*
>   * Remap block n of partition p to block n+start(p) of the disk.
>   */
> @@ -2141,7 +2148,7 @@ generic_make_request_checks(struct bio *bio)
>  	if ((bio->bi_opf & REQ_NOWAIT) && !queue_is_rq_based(q))
>  		goto not_supported;
>  
> -	if (should_fail_request(&bio->bi_disk->part0, bio->bi_iter.bi_size))
> +	if (should_fail_bio(bio))
>  		goto end_io;
>  
>  	if (blk_partition_remap(bio))
> -- 
> 2.14.1
> 

Re: [PATCH] block: Add should_fail_bio() for bpf error injection

[Jens Axboe]

On 2/6/18 12:27 PM, Omar Sandoval wrote:
> On Wed, Jan 24, 2018 at 03:22:58PM -0800, Howard McLauchlan wrote:
>> The classic error injection mechanism, should_fail_request() does not
>> support use cases where more information is required (from the entire
>> struct bio, for example).
>>
>> To that end, this patch introduces should_fail_bio(), which calls
>> should_fail_request() under the hood but provides a convenient
>> place for kprobes to hook into if they require the entire struct bio.
>> This patch also replaces some existing calls to should_fail_request()
>> with should_fail_bio() with no degradation in performance.
> 
> Reviewed-by: Omar Sandoval <osandov@fb.com>
> 
> Jens, can we pick this up for 4.16? The necessary BPF support has been
> merged into Linus' tree.

I don't see why not - but the patch no longer applies. Howard, can you
respin it?

-- 
Jens Axboe