Skip to contentSkip to conversation

zram: fix incorrectly stat with failed_reads

PATCH Superseded Chao Yu 3 participants 5 messages Aug 11, 2014 – Aug 13, 2014

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Chao Yu

author
#0 lore raw

PATCH 1/1 zram: fix incorrectly stat with failed_reads

Since we allocate a temporary buffer in zram_bvec_read to handle partial page operations in this commit 924bd88d703e53d30f393fac6117f8f1bc79aab6 (Staging:
zram: allow partial page operations), our ->failed_reads value may be incorrect as we do not increase its value when failed to allocate the temporary buffer.

Let's fix this issue and correct the annotation of failed_reads.

---
 drivers/block/zram/zram_drv.c | 1 +
 drivers/block/zram/zram_drv.h | 2 +-
 2 files changed, 2 insertions(+), 1 deletion(-)
drivers/block/zram/zram_drv.c +1 -0
index dfa4024..bf8ea1b 100644
@@ -413,6 +413,7 @@ static int zram_bvec_read(struct zram *zram, struct bio_vec *bvec,
 
 	if (!uncmem) {
 		pr_info("Unable to allocate temp memory\n");
+		atomic64_inc(&zram->stats.failed_reads);
 		ret = -ENOMEM;
 		goto out_cleanup;
 	}
drivers/block/zram/zram_drv.h +1 -1
index 5b0afde..e0f725c 100644
@@ -84,7 +84,7 @@ struct zram_stats {
 	atomic64_t compr_data_size;	/* compressed size of pages stored */
 	atomic64_t num_reads;	/* failed + successful */
 	atomic64_t num_writes;	/* --do-- */
-	atomic64_t failed_reads;	/* should NEVER! happen */
+	atomic64_t failed_reads;	/* can happen when memory is too low */
 	atomic64_t failed_writes;	/* can happen when memory is too low */
 	atomic64_t invalid_io;	/* non-page-aligned I/O requests */
 	atomic64_t notify_free;	/* no. of swap slot free notifications */
Signature · 1 line
2.0.1.474.g72c7794
maintainer
#1 lore raw
Quoted 6 lines · Chao Yu

Since we allocate a temporary buffer in zram_bvec_read to handle partial page operations in this commit 924bd88d703e53d30f393fac6117f8f1bc79aab6 (Staging:
zram: allow partial page operations), our ->failed_reads value may be incorrect as we do not increase its value when failed to allocate the temporary buffer.

Let's fix this issue and correct the annotation of failed_reads.

Good catch.
Just out of curiosity.

Did you catch it by code review or real workload?

Quoted 34 lines
---
 drivers/block/zram/zram_drv.c | 1 +
 drivers/block/zram/zram_drv.h | 2 +-
 2 files changed, 2 insertions(+), 1 deletion(-)
drivers/block/zram/zram_drv.c +1 -0
index dfa4024..bf8ea1b 100644
@@ -413,6 +413,7 @@ static int zram_bvec_read(struct zram *zram, struct bio_vec *bvec,
 
 	if (!uncmem) {
 		pr_info("Unable to allocate temp memory\n");
+		atomic64_inc(&zram->stats.failed_reads);
 		ret = -ENOMEM;
 		goto out_cleanup;
 	}
drivers/block/zram/zram_drv.h +1 -1
index 5b0afde..e0f725c 100644
@@ -84,7 +84,7 @@ struct zram_stats {
 	atomic64_t compr_data_size;	/* compressed size of pages stored */
 	atomic64_t num_reads;	/* failed + successful */
 	atomic64_t num_writes;	/* --do-- */
-	atomic64_t failed_reads;	/* should NEVER! happen */
+	atomic64_t failed_reads;	/* can happen when memory is too low */
 	atomic64_t failed_writes;	/* can happen when memory is too low */
 	atomic64_t invalid_io;	/* non-page-aligned I/O requests */
 	atomic64_t notify_free;	/* no. of swap slot free notifications */
Signature · 1 line
2.0.1.474.g72c7794

How abouting moving failed_reads/writes in zram_bvec_rw?

int zram_bvec_rw(..)

{
        if (rw == READ) {
                atomic64_inc(num_reads);
                ret = zram_bvec_read(xxx);
        } else {
                atomic64_inc(&num_writes);
                ret = zram_bvec_write(xxx);
        }

        if (unlikely(ret)) {
                if (rw == READ)
                        atomic64_inc(failed_reads);
                else
                        atomic64_inc(failed_writes);
        }
}
Quoted 7 lines

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Signature · 2 lines
Kind regards,
Minchan Kim
author
#2 lore raw

Hi Minchan,

Quoted 14 lines

-----Original Message-----
From: Minchan Kim [mailto:[email protected]]
Sent: Tuesday, August 12, 2014 3:38 PM
To: Chao Yu
Cc: [email protected]; [email protected]; Jerome Marchand; Sergey Senozhatsky; Andrew Morton
Subject: Re: [PATCH] zram: fix incorrectly stat with failed_reads

Quoted 5 lines · Chao Yu

Since we allocate a temporary buffer in zram_bvec_read to handle partial page operations in this commit 924bd88d703e53d30f393fac6117f8f1bc79aab6 (Staging:
zram: allow partial page operations), our ->failed_reads value may be incorrect as we do not increase its value when failed to allocate the temporary buffer.

Quoted 6 lines
Quoted 1 line

Let's fix this issue and correct the annotation of failed_reads.

Good catch.
Just out of curiosity.

Did you catch it by code review or real workload?

I found this one by code review.

Quoted 29 lines
Quoted 29 lines
---
 drivers/block/zram/zram_drv.c | 1 +
 drivers/block/zram/zram_drv.h | 2 +-
 2 files changed, 2 insertions(+), 1 deletion(-)
drivers/block/zram/zram_drv.c +1 -0
index dfa4024..bf8ea1b 100644
@@ -413,6 +413,7 @@ static int zram_bvec_read(struct zram *zram, struct bio_vec *bvec,
 
 	if (!uncmem) {
 		pr_info("Unable to allocate temp memory\n");
+		atomic64_inc(&zram->stats.failed_reads);
 		ret = -ENOMEM;
 		goto out_cleanup;
 	}
drivers/block/zram/zram_drv.h +0 -1
index 5b0afde..e0f725c 100644
@@ -84,7 +84,7 @@ struct zram_stats {
 	atomic64_t compr_data_size;	/* compressed size of pages stored */
 	atomic64_t num_reads;	/* failed + successful */
 	atomic64_t num_writes;	/* --do-- */
-	atomic64_t failed_reads;	/* should NEVER! happen */
Quoted 8 lines
Quoted 6 lines
drivers/block/zram/zram_drv.h +1 -1
+	atomic64_t failed_reads;	/* can happen when memory is too low */
 	atomic64_t failed_writes;	/* can happen when memory is too low */
 	atomic64_t invalid_io;	/* non-page-aligned I/O requests */
 	atomic64_t notify_free;	/* no. of swap slot free notifications */
--

2.0.1.474.g72c7794

How abouting moving failed_reads/writes in zram_bvec_rw?

I think this implementation of yours is pretty clean and simple because this one hides every error paths inside zram_bvec_{read,write}, so we do not need to consider covering the stat code to everywhere of inside.

IMHO, your implementation also fixed the issue in following call chain:
->zram_bvec_rw
->zram_bvec_write
->zram_decompress_page
->atomic64_inc(failed_reads) ---at perspective of user, this kind of
failure should be regarded as the write one's, because user pay fewer attention to inner error of zram.

Please correct me if I'm wrong, and thanks for your review and suggestion. :)

Regards,
Yu

Quoted 30 lines

int zram_bvec_rw(..)

{
        if (rw == READ) {
                atomic64_inc(num_reads);
                ret = zram_bvec_read(xxx);
        } else {
                atomic64_inc(&num_writes);
                ret = zram_bvec_write(xxx);
        }

        if (unlikely(ret)) {
                if (rw == READ)
                        atomic64_inc(failed_reads);
                else
                        atomic64_inc(failed_writes);
        }
}
Quoted 7 lines

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

--
Kind regards,
Minchan Kim

#3 lore raw
In reply to Minchan Kim
Quoted 51 lines · Minchan Kim
Quoted 6 lines · Chao Yu

Since we allocate a temporary buffer in zram_bvec_read to handle partial page operations in this commit 924bd88d703e53d30f393fac6117f8f1bc79aab6 (Staging:
zram: allow partial page operations), our ->failed_reads value may be incorrect as we do not increase its value when failed to allocate the temporary buffer.

Let's fix this issue and correct the annotation of failed_reads.

Good catch.
Just out of curiosity.

Did you catch it by code review or real workload?

Quoted 34 lines
---
 drivers/block/zram/zram_drv.c | 1 +
 drivers/block/zram/zram_drv.h | 2 +-
 2 files changed, 2 insertions(+), 1 deletion(-)
drivers/block/zram/zram_drv.c +1 -0
index dfa4024..bf8ea1b 100644
@@ -413,6 +413,7 @@ static int zram_bvec_read(struct zram *zram, struct bio_vec *bvec,
 
 	if (!uncmem) {
 		pr_info("Unable to allocate temp memory\n");
+		atomic64_inc(&zram->stats.failed_reads);
 		ret = -ENOMEM;
 		goto out_cleanup;
 	}
drivers/block/zram/zram_drv.h +1 -1
index 5b0afde..e0f725c 100644
@@ -84,7 +84,7 @@ struct zram_stats {
 	atomic64_t compr_data_size;	/* compressed size of pages stored */
 	atomic64_t num_reads;	/* failed + successful */
 	atomic64_t num_writes;	/* --do-- */
-	atomic64_t failed_reads;	/* should NEVER! happen */
+	atomic64_t failed_reads;	/* can happen when memory is too low */
 	atomic64_t failed_writes;	/* can happen when memory is too low */
 	atomic64_t invalid_io;	/* non-page-aligned I/O requests */
 	atomic64_t notify_free;	/* no. of swap slot free notifications */
Signature · 1 line
2.0.1.474.g72c7794

How abouting moving failed_reads/writes in zram_bvec_rw?

int zram_bvec_rw(..)

Quoted 17 lines
{
        if (rw == READ) {
                atomic64_inc(num_reads);
                ret = zram_bvec_read(xxx);
        } else {
                atomic64_inc(&num_writes);
                ret = zram_bvec_write(xxx);
        }

        if (unlikely(ret)) {
                if (rw == READ)
                        atomic64_inc(failed_reads);
                else
                        atomic64_inc(failed_writes);
        }
}

looks good. these stats were supposed to be here.

	-ss
Quoted 12 lines
Quoted 7 lines

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Signature · 2 lines
Kind regards,
Minchan Kim
author
#4 lore raw
In reply to Minchan Kim

Hi Minchan,

Quoted 15 lines

-----Original Message-----
From: Minchan Kim [mailto:[email protected]]
Sent: Tuesday, August 12, 2014 3:38 PM
To: Chao Yu
Cc: [email protected]; [email protected]; Jerome Marchand; Sergey Senozhatsky; Andrew Morton
Subject: Re: [PATCH] zram: fix incorrectly stat with failed_reads

Quoted 6 lines · Chao Yu

Since we allocate a temporary buffer in zram_bvec_read to handle partial page operations in this commit 924bd88d703e53d30f393fac6117f8f1bc79aab6 (Staging:
zram: allow partial page operations), our ->failed_reads value may be incorrect as we do not increase its value when failed to allocate the temporary buffer.

Let's fix this issue and correct the annotation of failed_reads.

[snip]

Quoted 19 lines

How abouting moving failed_reads/writes in zram_bvec_rw?

int zram_bvec_rw(..)

{
        if (rw == READ) {
                atomic64_inc(num_reads);
                ret = zram_bvec_read(xxx);
        } else {
                atomic64_inc(&num_writes);
                ret = zram_bvec_write(xxx);
        }

        if (unlikely(ret)) {
                if (rw == READ)
                        atomic64_inc(failed_reads);
                else
                        atomic64_inc(failed_writes);
        }
}

I will send a v2 patch base on above codes of yours, please help to review the following new patch.

Quoted 12 lines
Quoted 8 lines

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

--
Kind regards,
Minchan Kim