All of lore.kernel.org
 help / color / mirror / Atom feed
From: Greg Kurz <groug@kaod.org>
To: Christian Schoenebeck <qemu_oss@crudebyte.com>
Cc: qemu-devel@nongnu.org
Subject: Re: [PATCH v4 07/11] tests/virtio-9p: failing splitted readdir test
Date: Wed, 22 Jan 2020 23:59:54 +0100	[thread overview]
Message-ID: <20200122235954.1305faab@bahia.lan> (raw)
In-Reply-To: <4dc3706db1f033d922e54af8c74a81211de8b79f.1579567020.git.qemu_oss@crudebyte.com>

On Tue, 21 Jan 2020 01:17:35 +0100
Christian Schoenebeck <qemu_oss@crudebyte.com> wrote:

> This patch is not intended to be merged. It resembles
> an issue (with debug messages) where the splitted
> readdir test fails because server is interrupted with
> transport error "Failed to decode VirtFS request type 40",

Ok. When we send a new request, we call:

uint32_t qvirtqueue_add(QTestState *qts, QVirtQueue *vq, uint64_t data,
                        uint32_t len, bool write, bool next)
{
    uint16_t flags = 0;
    vq->num_free--;

[...]

    return vq->free_head++; /* Return and increase, in this order */
}

where vq->num_free is the number of available buffers (aka. requests) in
the vq and vq->free_head the index of the next available buffer. The size
of the vq of the virtio-9p device is MAX_REQ (128) buffers. The driver
is very simple and doesn't care to handle the scenario of a full vq,
ie, num_free == 0 and free_head is past the vq->desc[] array. It seems
that count=128 generates enough extra requests to reach the end of the
vq. Hence the "decode" error you get. Maybe an assert(vq->num_free) in
qvirtqueue_add() would make that more clear ?

Not sure it is worth to address this limitation though. Especially since
count=128 isn't really a recommended choice in the first place. It has
more chances to cause a disconnect if the server needs to return a longer
file name (which is expected since most fs have 255 character long file
names).

> which BTW fails both with the unoptimized and with the
> optimized 9p readdir code.
> 

Yes, this is the client's fault.

> Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
> ---
>  tests/qtest/virtio-9p-test.c | 9 ++++++++-
>  1 file changed, 8 insertions(+), 1 deletion(-)
> 
> diff --git a/tests/qtest/virtio-9p-test.c b/tests/qtest/virtio-9p-test.c
> index 8b0d94546e..e47b286340 100644
> --- a/tests/qtest/virtio-9p-test.c
> +++ b/tests/qtest/virtio-9p-test.c
> @@ -647,13 +647,14 @@ static void fs_readdir_split(void *obj, void *data, QGuestAllocator *t_alloc)
>      int fid;
>      uint64_t offset;
>      /* the Treaddir 'count' parameter values to be tested */
> -    const uint32_t vcount[] = { 512, 256 };
> +    const uint32_t vcount[] = { 512, 256, 128 };
>      const int nvcount = sizeof(vcount) / sizeof(uint32_t);
>  
>      fs_attach(v9p, NULL, t_alloc);
>  
>      /* iterate over all 'count' parameter values to be tested with Treaddir */
>      for (subtest = 0; subtest < nvcount; ++subtest) {
> +        printf("\nsubtest[%d] with count=%d\n", subtest, vcount[subtest]);
>          fid = subtest + 1;
>          offset = 0;
>          entries = NULL;
> @@ -674,12 +675,16 @@ static void fs_readdir_split(void *obj, void *data, QGuestAllocator *t_alloc)
>           * entries
>           */
>          while (true) {
> +            printf("\toffset=%ld\n", offset);
>              npartialentries = 0;
>              partialentries = NULL;
>  
> +            printf("Treaddir fid=%d offset=%ld count=%d\n",
> +                   fid, offset, vcount[subtest]);
>              req = v9fs_treaddir(v9p, fid, offset, vcount[subtest], 0);
>              v9fs_req_wait_for_reply(req, NULL);
>              v9fs_rreaddir(req, &count, &npartialentries, &partialentries);
> +            printf("\t\tnpartial=%d nentries=%d\n", npartialentries, nentries);
>              if (npartialentries > 0 && partialentries) {
>                  if (!entries) {
>                      entries = partialentries;
> @@ -716,6 +721,8 @@ static void fs_readdir_split(void *obj, void *data, QGuestAllocator *t_alloc)
>          }
>  
>          v9fs_free_dirents(entries);
> +
> +        printf("PASSED subtest[%d]\n", subtest);
>      }
>  
>      g_free(wnames[0]);



  reply	other threads:[~2020-01-22 23:01 UTC|newest]

Thread overview: 39+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-01-21  0:36 [PATCH v4 00/11] 9pfs: readdir optimization Christian Schoenebeck
2020-01-20 22:29 ` [PATCH v4 01/11] tests/virtio-9p: add terminating null in v9fs_string_read() Christian Schoenebeck
2020-01-20 22:47 ` [PATCH v4 02/11] 9pfs: require msize >= 4096 Christian Schoenebeck
2020-01-20 23:50 ` [PATCH v4 03/11] 9pfs: validate count sent by client with T_readdir Christian Schoenebeck
2020-01-22 14:11   ` Greg Kurz
2020-01-22 14:26     ` Christian Schoenebeck
2020-01-21  0:01 ` [PATCH v4 04/11] hw/9pfs/9p-synth: added directory for readdir test Christian Schoenebeck
2020-01-21  0:12 ` [PATCH v4 05/11] tests/virtio-9p: added " Christian Schoenebeck
2020-01-22 19:56   ` Greg Kurz
2020-01-21  0:16 ` [PATCH v4 06/11] tests/virtio-9p: added splitted " Christian Schoenebeck
2020-01-22 21:14   ` Eric Blake
2020-01-22 21:29     ` Christian Schoenebeck
2020-01-23  6:59       ` Greg Kurz
2020-01-22 21:19   ` Greg Kurz
2020-01-22 22:36     ` Christian Schoenebeck
2020-01-23 10:30       ` Greg Kurz
2020-01-23 13:07         ` Christian Schoenebeck
2020-01-21  0:17 ` [PATCH v4 07/11] tests/virtio-9p: failing " Christian Schoenebeck
2020-01-22 22:59   ` Greg Kurz [this message]
2020-01-23 11:36     ` Christian Schoenebeck
2020-01-23 12:08       ` Greg Kurz
2020-01-21  0:23 ` [PATCH v4 08/11] 9pfs: readdir benchmark Christian Schoenebeck
2020-01-23 10:34   ` Greg Kurz
2020-01-23 13:20     ` Christian Schoenebeck
2020-01-21  0:26 ` [PATCH v4 09/11] hw/9pfs/9p-synth: avoid n-square issue in synth_readdir() Christian Schoenebeck
2020-01-23 11:13   ` Greg Kurz
2020-01-23 12:40     ` Christian Schoenebeck
2020-01-21  0:30 ` [PATCH v4 10/11] 9pfs: T_readdir latency optimization Christian Schoenebeck
2020-01-23 11:33   ` Greg Kurz
2020-01-23 12:57     ` Christian Schoenebeck
2020-03-09 14:09   ` Christian Schoenebeck
2020-03-09 15:42     ` Greg Kurz
2020-03-10 15:10       ` Christian Schoenebeck
2020-03-10 18:33         ` Greg Kurz
2020-03-11  1:18           ` Christian Schoenebeck
     [not found]             ` <20200311171408.3b3a2dfa@bahia.home>
2020-03-11 19:54               ` Christian Schoenebeck
2020-03-17 14:14                 ` Greg Kurz
2020-03-17 16:09                   ` Christian Schoenebeck
2020-01-21  0:32 ` [PATCH v4 11/11] hw/9pfs/9p.c: benchmark time on T_readdir request Christian Schoenebeck

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=20200122235954.1305faab@bahia.lan \
    --to=groug@kaod.org \
    --cc=qemu-devel@nongnu.org \
    --cc=qemu_oss@crudebyte.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.