Re: [PATCH] block: Invalidate the cache for a parent block-device

[Niels de Vos]

On 01/20/2012 01:58 AM, Mikulas Patocka wrote:
> This definitely doesn't belong to blkdev_issue_flush.
> 
> blkdev_issue_flush is called by filesystems each time they want to 
> synchronize hardware disk cache (for example, when committing a journal).

Ay, that sounds bad.

> The patch may cause serious performance regressions (each time a 
> filesystem commits journal, the patch causes it to walk all buffers in the 
> whole-disk device).
> 
> You should put this code into fsync_bdev (so that it is called only on 
> fsync or BLKFLSBUF) and not to blkdev_issue_flush.

Hmm, I actually thought about that earlier, but decided that
blkdev_issue_flush() was nicer. Looking at it again, I'm not sure why I
thought that.

I'll post a 2nd version soon.

Thanks,
Niels


> Mikulas
> 
>> Executing a BLKFLSBUF-ioctl on a partition flushes the caches for that
>> partition but reading data through the parent device will still return
>> the old cached data.
>>
>> The cache for the block-device is not synced if the block-device is kept
>> open (due to a mounted partition, for example). Only when all users for
>> the disk have exited, the cache for the disk is made consistent again.
>>
>> Calling invalidate_bdev() on the parent block-device in case
>> blkdev_issue_flush() was called for a partition fixes this.
>>
>> The problem can be worked around by forcing the caches to be flushed
>> with either
>>         # blockdev --flushbufs ${dev_disk}
>> or
>>         # echo 3 > /proc/sys/vm/drop_caches
>>
>> CC: Bryn M. Reeves <bmr <at> redhat.com>
>> Signed-off-by: Niels de Vos <ndevos <at> redhat.com>
>> ---
>>  block/blk-flush.c |    3 +++
>>  1 files changed, 3 insertions(+), 0 deletions(-)
>>
>> diff --git a/block/blk-flush.c b/block/blk-flush.c
>> index 720ad60..e876f8e 100644
>> --- a/block/blk-flush.c
>> +++ b/block/blk-flush.c
>> @@ -448,6 +448,9 @@ int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_
>>
>>         if (!bio_flagged(bio, BIO_UPTODATE))
>>                 ret = -EIO;
>> +       else if (bdev != bdev->bd_contains)
>> +               /* invalidate parent block_device */
>> +               invalidate_bdev(bdev->bd_contains);
>>
>>         bio_put(bio);
>>         return ret;
>> --
>> 1.7.6.5
>>

-- 
Niels de Vos
Software Maintenance Engineer
Support Engineering Group
Red Hat UK, Ltd.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/