[Qemu-devel] Re: [PATCH 5/5] drive_open: Add invalidate option for block devices

[Juan Quintela <quintela@redhat.com>]

Blue Swirl <blauwirbel@gmail.com> wrote:
> On Tue, Jan 4, 2011 at 2:33 PM, Juan Quintela <quintela@redhat.com> wrote:
>> Linux allows to invalidate block devices.  This is needed for the incoming
>> migration part.
>>
>> Signed-off-by: Juan Quintela <quintela@redhat.com>
>> ---
>>  block.h           |    2 ++
>>  block/raw-posix.c |   24 ++++++++++++++++++++++++
>>  blockdev.c        |    9 +++++----
>>  3 files changed, 31 insertions(+), 4 deletions(-)
>>
>> diff --git a/block.h b/block.h
>> index f923add..5ac96a5 100644
>> --- a/block.h
>> +++ b/block.h
>> @@ -34,6 +34,8 @@ typedef struct QEMUSnapshotInfo {
>>  #define BDRV_O_NATIVE_AIO  0x0080 /* use native AIO instead of the thread pool */
>>  #define BDRV_O_NO_BACKING  0x0100 /* don't open the backing file */
>>  #define BDRV_O_NO_FLUSH    0x0200 /* disable flushing on this disk */
>> +#define BDRV_O_INVALIDATE  0x0400 /* invalidate buffer cache for this device.
>> +                                     re-read things from server */
>>
>>  #define BDRV_O_CACHE_MASK  (BDRV_O_NOCACHE | BDRV_O_CACHE_WB | BDRV_O_NO_FLUSH)
>>
>> diff --git a/block/raw-posix.c b/block/raw-posix.c
>> index 6b72470..9439cf1 100644
>> --- a/block/raw-posix.c
>> +++ b/block/raw-posix.c
>> @@ -51,6 +51,7 @@
>>  #include <sys/param.h>
>>  #include <linux/cdrom.h>
>>  #include <linux/fd.h>
>> +#include <linux/fs.h>
>>  #endif
>>  #if defined (__FreeBSD__) || defined(__FreeBSD_kernel__)
>>  #include <signal.h>
>> @@ -168,6 +169,29 @@ static int raw_open_common(BlockDriverState *bs, const char *filename,
>>     s->fd = fd;
>>     s->aligned_buf = NULL;
>>
>> +#ifdef __linux__
>> +    if ((bdrv_flags & BDRV_O_INVALIDATE)) {
>> +        struct stat buf;
>> +        int res;
>> +
>> +        res = fstat(fd, &buf);
>> +
>> +        if (res < 0) {
>> +            return -errno;
>> +        }
>> +
>> +        if (S_ISBLK(buf.st_mode)) {
>> +            printf("we are in a block device: %s\n", filename);
>
> Leftover debugging?

ouch, yes.

thanks for the spotting.

Later, Juan.