qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Kevin Wolf <kwolf@redhat.com>
To: Paolo Bonzini <pbonzini@redhat.com>
Cc: pl@kamp.de, qemu-devel@nongnu.org, stefanha@redhat.com
Subject: Re: [Qemu-devel] [PATCH v3 13/19] qemu-img: add a "map" subcommand
Date: Tue, 30 Jul 2013 17:13:53 +0200	[thread overview]
Message-ID: <20130730151353.GF2475@dhcp-200-207.str.redhat.com> (raw)
In-Reply-To: <1374762197-7261-14-git-send-email-pbonzini@redhat.com>

Am 25.07.2013 um 16:23 hat Paolo Bonzini geschrieben:
> This command dumps the metadata of an entire chain, in either tabular or JSON
> format.
> 
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Hm, we have a 'map' command in qemu-io, which isn't exactly the same,
but then not much different either.

Depending on the use cases, should we move this to qemu-io, should we
remove the old function from qemu-io, or should we really keep both?

> diff --git a/qemu-img.c b/qemu-img.c
> index c5c8ebc..b28d388 100644
> --- a/qemu-img.c
> +++ b/qemu-img.c
> @@ -1768,6 +1768,192 @@ static int img_info(int argc, char **argv)
>      return 0;
>  }
>  
> +
> +typedef struct MapEntry {
> +    int flags;
> +    int depth;
> +    int64_t start;
> +    int64_t length;
> +    int64_t offset;
> +} MapEntry;
> +
> +static void dump_map_entry(OutputFormat output_format, MapEntry *e,
> +                           MapEntry *next)
> +{
> +    switch (output_format) {
> +    case OFORMAT_HUMAN:
> +        if ((e->flags & BDRV_BLOCK_DATA) &&
> +            !(e->flags & BDRV_BLOCK_OFFSET_VALID)) {
> +            error_report("File contains external, encrypted or compressed clusters.");
> +            exit(1);
> +        }
> +        if ((e->flags & (BDRV_BLOCK_DATA|BDRV_BLOCK_ZERO)) == BDRV_BLOCK_DATA) {
> +            printf("%"PRId64" %"PRId64" %d %"PRId64"\n",
> +                   e->start, e->length, e->depth, e->offset);

Is this really human-readable output?

> +        }
> +        /* This format ignores the distinction between 0, ZERO and ZERO|DATA.
> +         * Modify the flags here to allow more coalescing.
> +         */
> +        if (next &&
> +            (next->flags & (BDRV_BLOCK_DATA|BDRV_BLOCK_ZERO)) != BDRV_BLOCK_DATA) {
> +            next->flags &= ~BDRV_BLOCK_DATA;
> +            next->flags |= BDRV_BLOCK_ZERO;
> +        }
> +        break;
> +    case OFORMAT_JSON:
> +        printf("%s{ 'start': %"PRId64", 'length': %"PRId64", 'depth': %d, "
> +               "'zero': %s, 'data': %s",
> +               (e->start == 0 ? "[" : ",\n"),
> +               e->start, e->length, e->depth,
> +               (e->flags & BDRV_BLOCK_ZERO) ? "true" : "false",
> +               (e->flags & BDRV_BLOCK_DATA) ? "true" : "false");

Correct JSON uses double quotes.

> +        if (e->flags & BDRV_BLOCK_OFFSET_VALID) {
> +            printf(", 'offset': %"PRId64"", e->offset);
> +        }
> +        putchar('}');
> +
> +        if (!next) {
> +            printf("]\n");
> +        }
> +        break;
> +    }
> +}
> +
> +static int64_t get_block_status(BlockDriverState *bs, int64_t sector_num,
> +                                int *nb_sectors, int *depth)
> +{
> +    int64_t ret;
> +
> +    /* As an optimization, we could cache the current range of unallocated
> +     * clusters in each file of the chain, and avoid querying the same
> +     * range repeatedly.
> +     */
> +
> +    *depth = 0;
> +    for (;;) {
> +        int orig_nb_sectors = *nb_sectors;
> +
> +        ret = bdrv_get_block_status(bs, sector_num, *nb_sectors, nb_sectors);
> +        if (ret < 0) {
> +            return ret;
> +        }
> +        if (ret & (BDRV_BLOCK_ZERO|BDRV_BLOCK_DATA)) {
> +            return ret;
> +        }
> +        if (!*nb_sectors) {
> +            /* Beyond the end of this image.  The extra data is read as zeroes.
> +             * We're in the range of the BlockDriverState above this one, so
> +             * adjust depth.
> +             */
> +            *nb_sectors = orig_nb_sectors;
> +            (*depth)--;
> +            return BDRV_BLOCK_ZERO;

If you implement my suggestion that bdrv_co_get_block_status() returns
BDRV_BLOCK_ZERO instead of 0 if the image has no backing file, it might
also make sense to put this check there and return BDRV_BLOCK_ZERO even
if it has a backing file, but we're after its end.

> +        }
> +
> +        bs = bs->backing_hd;
> +        if (bs == NULL) {
> +            return 0;
> +        }
> +
> +        (*depth)++;
> +    }
> +}
> +
> +static int img_map(int argc, char **argv)
> +{
> +    int c;
> +    OutputFormat output_format = OFORMAT_HUMAN;
> +    BlockDriverState *bs;
> +    const char *filename, *fmt, *output;
> +    int64_t length;
> +    MapEntry curr = { .length = 0 }, next;
> +
> +    fmt = NULL;
> +    output = NULL;
> +    for (;;) {
> +        int option_index = 0;
> +        static const struct option long_options[] = {
> +            {"help", no_argument, 0, 'h'},
> +            {"format", required_argument, 0, 'f'},
> +            {"output", required_argument, 0, OPTION_OUTPUT},
> +            {0, 0, 0, 0}
> +        };
> +        c = getopt_long(argc, argv, "f:h",
> +                        long_options, &option_index);
> +        if (c == -1) {
> +            break;
> +        }
> +        switch (c) {
> +        case '?':
> +        case 'h':
> +            help();
> +            break;
> +        case 'f':
> +            fmt = optarg;
> +            break;
> +        case OPTION_OUTPUT:
> +            output = optarg;
> +            break;
> +        }
> +    }
> +    if (optind >= argc) {
> +        help();
> +    }
> +    filename = argv[optind++];
> +
> +    if (output && !strcmp(output, "json")) {
> +        output_format = OFORMAT_JSON;
> +    } else if (output && !strcmp(output, "human")) {
> +        output_format = OFORMAT_HUMAN;
> +    } else if (output) {
> +        error_report("--output must be used with human or json as argument.");
> +        return 1;
> +    }
> +
> +    bs = bdrv_new_open(filename, fmt, BDRV_O_FLAGS, true, false);
> +    if (!bs) {
> +        return 1;
> +    }
> +
> +    length = bdrv_getlength(bs);

bdrv_getlength() can fail.

> +    while (curr.start + curr.length < length) {
> +        int64_t nsectors_left, ret;
> +        int64_t sector_num;
> +        int n, depth, flags;
> +
> +        /* Probe up to 1 G at a time.  */
> +        sector_num = (curr.start + curr.length) >> BDRV_SECTOR_BITS;
> +        nsectors_left = DIV_ROUND_UP(length, BDRV_SECTOR_SIZE) - sector_num;
> +        n = MIN(1 << (30 - BDRV_SECTOR_BITS), nsectors_left);
> +        ret = get_block_status(bs, sector_num, &n, &depth);
> +
> +        if (ret < 0) {
> +            error_report("Could not read file metadata: %s", strerror(-ret));
> +            return 1;

This leaks bs.

> +        }
> +
> +        flags = ret & ~BDRV_BLOCK_OFFSET_MASK;
> +        ret &= BDRV_BLOCK_OFFSET_MASK;
> +        if (curr.length == 0 || curr.flags != flags || curr.depth != depth ||
> +            ((curr.flags & BDRV_BLOCK_OFFSET_VALID) &&
> +             curr.offset + curr.length != ret)) {
> +            next.flags = flags;
> +            next.depth = depth;
> +            next.start = sector_num << BDRV_SECTOR_BITS;
> +            next.offset = ret;
> +            next.length = 0;
> +            if (curr.length > 0) {
> +                dump_map_entry(output_format, &curr, &next);
> +            }
> +            curr = next;
> +        }
> +        curr.length += n << BDRV_SECTOR_BITS;
> +    }
> +
> +    dump_map_entry(output_format, &curr, NULL);
> +    return 0;
> +}
> +
>  #define SNAPSHOT_LIST   1
>  #define SNAPSHOT_CREATE 2
>  #define SNAPSHOT_APPLY  3

Kevin

  reply	other threads:[~2013-07-30 15:14 UTC|newest]

Thread overview: 56+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-07-25 14:22 [Qemu-devel] [PATCH v3 00/19] Add qemu-img subcommand to dump file metadata Paolo Bonzini
2013-07-25 14:22 ` [Qemu-devel] [PATCH v3 01/19] cow: make reads go at a decent speed Paolo Bonzini
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 02/19] cow: make writes go at a less indecent speed Paolo Bonzini
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 03/19] cow: do not call bdrv_co_is_allocated Paolo Bonzini
2013-07-29 13:22   ` Kevin Wolf
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 04/19] block: update bs->total_sectors on writes Paolo Bonzini
2013-07-29 13:13   ` Kevin Wolf
2013-07-29 13:47     ` Paolo Bonzini
2013-07-29 14:10       ` Kevin Wolf
2013-07-29 14:18         ` Paolo Bonzini
2013-08-02  7:05           ` Peter Lieven
2013-08-17  6:27             ` Paolo Bonzini
2013-09-02  7:12               ` Peter Lieven
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 05/19] block: make bdrv_co_is_allocated static Paolo Bonzini
2013-07-29 13:21   ` Kevin Wolf
2013-07-29 13:56     ` Paolo Bonzini
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 06/19] block: remove bdrv_is_allocated_above/bdrv_co_is_allocated_above distinction Paolo Bonzini
2013-07-29 13:34   ` Kevin Wolf
2013-07-29 13:59     ` Paolo Bonzini
2013-07-29 14:15       ` Kevin Wolf
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 07/19] block: expect errors from bdrv_co_is_allocated Paolo Bonzini
2013-07-29 13:43   ` Kevin Wolf
2013-07-29 14:03     ` Paolo Bonzini
2013-07-29 14:17       ` Kevin Wolf
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 08/19] qemu-img: always probe the input image for allocated sectors Paolo Bonzini
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 09/19] block: make bdrv_has_zero_init return false for copy-on-write-images Paolo Bonzini
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 10/19] block: introduce bdrv_get_block_status API Paolo Bonzini
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 11/19] block: define get_block_status return value Paolo Bonzini
2013-07-30 14:14   ` Kevin Wolf
2013-07-30 14:19     ` Paolo Bonzini
2013-07-30 14:26       ` Kevin Wolf
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 12/19] block: return get_block_status data and flags for formats Paolo Bonzini
2013-07-30 14:40   ` Kevin Wolf
2013-07-30 15:15     ` Paolo Bonzini
2013-07-30 15:23       ` Kevin Wolf
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 13/19] qemu-img: add a "map" subcommand Paolo Bonzini
2013-07-30 15:13   ` Kevin Wolf [this message]
2013-07-30 15:22     ` Paolo Bonzini
2013-07-30 15:30       ` Kevin Wolf
2013-07-31  8:57       ` Kevin Wolf
2013-07-31 12:13         ` Paolo Bonzini
2013-07-31 13:26           ` Kevin Wolf
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 14/19] docs, qapi: document qemu-img map Paolo Bonzini
2013-07-30 15:48   ` Eric Blake
2013-07-30 15:54     ` Paolo Bonzini
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 15/19] block: use bdrv_has_zero_init to return BDRV_BLOCK_ZERO Paolo Bonzini
2013-07-31  9:16   ` Kevin Wolf
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 16/19] raw-posix: return get_block_status data and flags Paolo Bonzini
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 17/19] raw-posix: detect XFS unwritten extents Paolo Bonzini
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 18/19] block: add default get_block_status implementation for protocols Paolo Bonzini
2013-07-31  9:12   ` Kevin Wolf
2013-07-31 12:49     ` Paolo Bonzini
2013-07-25 14:23 ` [Qemu-devel] [PATCH v3 19/19] block: look for zero blocks in bs->file Paolo Bonzini
2013-07-31  9:21   ` Kevin Wolf
2013-07-31 12:53     ` Paolo Bonzini
2013-08-16 15:30 ` [Qemu-devel] [PATCH v3 00/19] Add qemu-img subcommand to dump file metadata Stefan Hajnoczi

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20130730151353.GF2475@dhcp-200-207.str.redhat.com \
    --to=kwolf@redhat.com \
    --cc=pbonzini@redhat.com \
    --cc=pl@kamp.de \
    --cc=qemu-devel@nongnu.org \
    --cc=stefanha@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).