[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-block] [RFC 2/2] block/file-posix: verify page cache is not us
From: |
Dr. David Alan Gilbert |
Subject: |
Re: [Qemu-block] [RFC 2/2] block/file-posix: verify page cache is not used |
Date: |
Thu, 19 Apr 2018 10:05:47 +0100 |
User-agent: |
Mutt/1.9.5 (2018-04-13) |
* Stefan Hajnoczi (address@hidden) wrote:
> This commit is for debugging only. Do not merge it.
>
> mincore(2) checks whether pages are resident. Use it to verify that
> page cache has been dropped.
>
> You can trigger a verification failure by mmapping the image file from
> another process and loading a byte from a page so that it becomes
> resident. bdrv_co_invalidate_cache() will fail while the process is
> alive.
It doesn't seem a bad diagnostic to keep in (with a switch to activate)
for when we're faced with some weird corruption on some weird storage
system.
Dave
> Signed-off-by: Stefan Hajnoczi <address@hidden>
> ---
> block/file-posix.c | 71
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 71 insertions(+)
>
> diff --git a/block/file-posix.c b/block/file-posix.c
> index df4f52919f..d3105269c6 100644
> --- a/block/file-posix.c
> +++ b/block/file-posix.c
> @@ -2236,6 +2236,75 @@ static int coroutine_fn
> raw_co_block_status(BlockDriverState *bs,
> return ret | BDRV_BLOCK_OFFSET_VALID;
> }
>
> +static bool is_mincore(void *addr, size_t length)
> +{
> + size_t vec_len = DIV_ROUND_UP(length, sysconf(_SC_PAGESIZE));
> + unsigned char *vec;
> + size_t i;
> + int ret;
> + bool incore = false;
> +
> + vec = g_malloc(vec_len);
> + ret = mincore(addr, length, vec);
> + if (ret < 0) {
> + incore = true;
> + goto out;
> + }
> +
> + for (i = 0; i < vec_len; i++) {
> + if (vec[i] & 0x1) {
> + incore = true;
> + break;
> + }
> + }
> +
> +out:
> + g_free(vec);
> + return incore;
> +}
> +
> +static void check_not_in_page_cache(BlockDriverState *bs, Error **errp)
> +{
> + const size_t WINDOW_SIZE = 128 * 1024 * 1024;
> + BDRVRawState *s = bs->opaque;
> + void *window = NULL;
> + size_t length = 0;
> + off_t end;
> + off_t offset;
> +
> + end = raw_getlength(bs);
> +
> + for (offset = 0; offset < end; offset += WINDOW_SIZE) {
> + void *new_window;
> + size_t new_length = MIN(end - offset, WINDOW_SIZE);
> +
> + if (new_length != length) {
> + munmap(window, length);
> + window = NULL;
> + length = 0;
> + }
> +
> + new_window = mmap(window, new_length, PROT_NONE, MAP_PRIVATE,
> + s->fd, offset);
> + if (new_window == MAP_FAILED) {
> + error_setg_errno(errp, errno, "mmap failed");
> + break;
> + }
> +
> + window = new_window;
> + length = new_length;
> +
> + if (is_mincore(window, length)) {
> + error_setg(errp, "page cache still in use!");
> + break;
> + }
> + }
> +
> + if (window) {
> + munmap(window, length);
> + }
> +}
> +
> static void coroutine_fn raw_co_invalidate_cache(BlockDriverState *bs,
> Error **errp)
> {
> @@ -2270,6 +2339,8 @@ static void coroutine_fn
> raw_co_invalidate_cache(BlockDriverState *bs,
> return;
> }
> #endif /* __linux__ */
> +
> + check_not_in_page_cache(bs, errp);
> }
>
> static coroutine_fn BlockAIOCB *raw_aio_pdiscard(BlockDriverState *bs,
> --
> 2.14.3
>
--
Dr. David Alan Gilbert / address@hidden / Manchester, UK
[Qemu-block] [RFC 2/2] block/file-posix: verify page cache is not used, Stefan Hajnoczi, 2018/04/19
- Re: [Qemu-block] [RFC 2/2] block/file-posix: verify page cache is not used,
Dr. David Alan Gilbert <=
- Re: [Qemu-block] [RFC 2/2] block/file-posix: verify page cache is not used, Stefan Hajnoczi, 2018/04/19
- Re: [Qemu-block] [RFC 2/2] block/file-posix: verify page cache is not used, Kevin Wolf, 2018/04/20
- Re: [Qemu-block] [RFC 2/2] block/file-posix: verify page cache is not used, Stefan Hajnoczi, 2018/04/24
- Re: [Qemu-block] [RFC 2/2] block/file-posix: verify page cache is not used, Kevin Wolf, 2018/04/24
- Re: [Qemu-block] [RFC 2/2] block/file-posix: verify page cache is not used, Stefan Hajnoczi, 2018/04/27
Re: [Qemu-block] [Qemu-devel] [RFC 0/2] block/file-posix: allow -drive cache.direct=off live migration, Eric Blake, 2018/04/19