From: Paolo Bonzini <pbonzini@redhat.com>
To: Jeff Cody <jcody@redhat.com>
Cc: rjones@redhat.com, qemu-devel@nongnu.org, qemu-block@nongnu.org,
qemu-stable@nongnu.org
Subject: Re: [Qemu-devel] [PATCH 3/7] curl: avoid recursive locking of BDRVCURLState mutex
Date: Fri, 12 May 2017 16:48:10 +0200 [thread overview]
Message-ID: <ca25db4d-0c55-50a6-7cb0-152a76e5a875@redhat.com> (raw)
In-Reply-To: <20170511205611.GC19824@localhost.localdomain>
On 11/05/2017 22:56, Jeff Cody wrote:
> On Wed, May 10, 2017 at 04:32:01PM +0200, Paolo Bonzini wrote:
>> The curl driver has a ugly hack where, if it cannot find an empty CURLState,
>> it just uses aio_poll to wait for one to be empty. This is probably
>> buggy when used together with dataplane, and the simplest way to fix it
>> is to use coroutines instead.
>>
>> A more immediate effect of the bug however is that it can cause a
>> recursive call to curl_readv_bh_cb and recursively taking the
>> BDRVCURLState mutex. This causes a deadlock.
>>
>> The fix is to unlock the mutex around aio_poll, but for cleanliness we
>> should also take the mutex around all calls to curl_init_state, even if
>> reaching the unlock/lock pair is impossible. The same is true for
>> curl_clean_state.
>>
>> Reported-by: Richard W.M. Jones <rjones@redhat.com>
>> Cc: jcody@redhat.com
>> Cc: qemu-stable@nongnu.org
>> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
>> ---
>> block/curl.c | 13 ++++++++++++-
>> 1 file changed, 12 insertions(+), 1 deletion(-)
>>
>> diff --git a/block/curl.c b/block/curl.c
>> index 9a00fdc28e..b18e79bf54 100644
>> --- a/block/curl.c
>> +++ b/block/curl.c
>> @@ -281,6 +281,7 @@ read_end:
>> return size * nmemb;
>> }
>>
>> +/* Called with s->mutex held. */
>> static int curl_find_buf(BDRVCURLState *s, size_t start, size_t len,
>> CURLAIOCB *acb)
>> {
>> @@ -453,6 +454,7 @@ static void curl_multi_timeout_do(void *arg)
>> #endif
>> }
>>
>> +/* Called with s->mutex held. */
>> static CURLState *curl_init_state(BlockDriverState *bs, BDRVCURLState *s)
>> {
>> CURLState *state = NULL;
>> @@ -471,7 +473,9 @@ static CURLState *curl_init_state(BlockDriverState *bs, BDRVCURLState *s)
>> break;
>> }
>> if (!state) {
>> + qemu_mutex_unlock(&s->mutex);
>> aio_poll(bdrv_get_aio_context(bs), true);
>> + qemu_mutex_lock(&s->mutex);
>> }
>> } while(!state);
>>
>> @@ -534,6 +538,7 @@ static CURLState *curl_init_state(BlockDriverState *bs, BDRVCURLState *s)
>> return state;
>> }
>>
>> +/* Called with s->mutex held. */
>> static void curl_clean_state(CURLState *s)
>> {
>> int j;
>> @@ -565,6 +570,7 @@ static void curl_detach_aio_context(BlockDriverState *bs)
>> BDRVCURLState *s = bs->opaque;
>> int i;
>>
>> + qemu_mutex_lock(&s->mutex);
>> for (i = 0; i < CURL_NUM_STATES; i++) {
>> if (s->states[i].in_use) {
>> curl_clean_state(&s->states[i]);
>> @@ -580,6 +586,7 @@ static void curl_detach_aio_context(BlockDriverState *bs)
>> curl_multi_cleanup(s->multi);
>> s->multi = NULL;
>> }
>> + qemu_mutex_unlock(&s->mutex);
>>
>> timer_del(&s->timer);
>> }
>> @@ -745,9 +752,12 @@ static int curl_open(BlockDriverState *bs, QDict *options, int flags,
>> }
>>
>> DPRINTF("CURL: Opening %s\n", file);
>> + qemu_mutex_init(&s->mutex);
>
> This mutex init is now done above possible returns on error, so we should
> call qemu_mutex_destroy() on errors after this point.
Ok, I'll wait for you to complete the review and send v3.
Paolo
next prev parent reply other threads:[~2017-05-12 14:48 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-05-10 14:31 [Qemu-devel] [PATCH v2 0/7] curl: locking cleanups/fixes, coroutine conversion, remove aio_poll Paolo Bonzini
2017-05-10 14:31 ` [Qemu-devel] [PATCH 1/7] curl: strengthen assertion in curl_clean_state Paolo Bonzini
2017-05-10 16:33 ` [Qemu-devel] [Qemu-block] " Max Reitz
2017-05-11 20:35 ` [Qemu-devel] " Jeff Cody
2017-05-10 14:32 ` [Qemu-devel] [PATCH 2/7] curl: never invoke callbacks with s->mutex held Paolo Bonzini
2017-05-10 16:33 ` [Qemu-devel] [Qemu-block] " Max Reitz
2017-05-11 20:40 ` [Qemu-devel] " Jeff Cody
2017-05-10 14:32 ` [Qemu-devel] [PATCH 3/7] curl: avoid recursive locking of BDRVCURLState mutex Paolo Bonzini
2017-05-10 16:38 ` [Qemu-devel] [Qemu-block] " Max Reitz
2017-05-11 20:56 ` [Qemu-devel] " Jeff Cody
2017-05-12 14:48 ` Paolo Bonzini [this message]
2017-05-10 14:32 ` [Qemu-devel] [PATCH 4/7] curl: split curl_find_state/curl_init_state Paolo Bonzini
2017-05-10 17:26 ` [Qemu-devel] [Qemu-block] " Max Reitz
2017-05-11 13:49 ` [Qemu-devel] " Paolo Bonzini
2017-05-12 21:38 ` Jeff Cody
2017-05-10 14:32 ` [Qemu-devel] [PATCH 5/7] curl: convert CURLAIOCB to byte values Paolo Bonzini
2017-05-10 17:36 ` [Qemu-devel] [Qemu-block] " Max Reitz
2017-05-10 18:37 ` Eric Blake
2017-05-12 21:38 ` [Qemu-devel] " Jeff Cody
2017-05-10 14:32 ` [Qemu-devel] [PATCH 6/7] curl: convert readv to coroutines Paolo Bonzini
2017-05-12 21:40 ` Jeff Cody
2017-05-10 14:32 ` [Qemu-devel] [PATCH 7/7] curl: do not do aio_poll when waiting for a free CURLState Paolo Bonzini
2017-05-10 17:54 ` [Qemu-devel] [Qemu-block] " Max Reitz
2017-05-12 21:41 ` [Qemu-devel] " Jeff Cody
2017-05-10 15:11 ` [Qemu-devel] [PATCH v2 0/7] curl: locking cleanups/fixes, coroutine conversion, remove aio_poll no-reply
2017-05-10 15:57 ` Richard W.M. Jones
2017-05-15 19:12 ` [Qemu-devel] [Qemu-block] " Max Reitz
2017-05-15 20:30 ` Richard W.M. Jones
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=ca25db4d-0c55-50a6-7cb0-152a76e5a875@redhat.com \
--to=pbonzini@redhat.com \
--cc=jcody@redhat.com \
--cc=qemu-block@nongnu.org \
--cc=qemu-devel@nongnu.org \
--cc=qemu-stable@nongnu.org \
--cc=rjones@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).