From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:54517) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1b7K5Z-0006fW-0e for qemu-devel@nongnu.org; Mon, 30 May 2016 06:06:28 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1b7K5S-0000Cr-7J for qemu-devel@nongnu.org; Mon, 30 May 2016 06:06:23 -0400 Date: Mon, 30 May 2016 12:06:10 +0200 From: Kevin Wolf Message-ID: <20160530100610.GD4436@noname.str.redhat.com> References: <413c4af8-ba30-bfbf-9a84-d1d92ddac0e4@redhat.com> <20160526083016.GD31052@ad.usersys.redhat.com> <4b26422f-9320-31a6-9dca-534bf0ea6086@redhat.com> <20160527003602.GG31052@ad.usersys.redhat.com> <20160527085559.GC5467@noname.redhat.com> <574BDCC4.8000704@kamp.de> <20160530082412.GB4436@noname.str.redhat.com> <574C0822.6040004@kamp.de> <20160530094719.GC4436@noname.str.redhat.com> <574C0D92.5090302@kamp.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <574C0D92.5090302@kamp.de> Subject: Re: [Qemu-devel] [PATCH V2] block/io: optimize bdrv_co_pwritev for small requests List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Peter Lieven Cc: Fam Zheng , Paolo Bonzini , qemu-block@nongnu.org, qemu-devel@nongnu.org, stefanha@redhat.com, mreitz@redhat.com Am 30.05.2016 um 11:53 hat Peter Lieven geschrieben: > Am 30.05.2016 um 11:47 schrieb Kevin Wolf: > >Am 30.05.2016 um 11:30 hat Peter Lieven geschrieben: > >>Am 30.05.2016 um 10:24 schrieb Kevin Wolf: > >>>Am 30.05.2016 um 08:25 hat Peter Lieven geschrieben: > >>>>Am 27.05.2016 um 10:55 schrieb Kevin Wolf: > >>>>>Am 27.05.2016 um 02:36 hat Fam Zheng geschrieben: > >>>>>>On Thu, 05/26 11:20, Paolo Bonzini wrote: > >>>>>>>On 26/05/2016 10:30, Fam Zheng wrote: > >>>>>>>>>>This doesn't look too wrong... Should the right sequence of events be > >>>>>>>>>>head/after_head or head/after_tail? It's probably simplest to just emit > >>>>>>>>>>all four events. > >>>>>>>>I've no idea. (That's why I leaned towards fixing the test case). > >>>>>>>Well, fixing the testcase means knowing what events should be emitted. > >>>>>>> > >>>>>>>QEMU with Peter's patch emits head/after_head. If the right one is > >>>>>>>head/after_tail, _both QEMU and the testcase_ need to be adjusted. Your > >>>>>>>patch keeps the backwards-compatible route. > >>>>>>Yes, I mean I was not very convinced in tweaking the events at all: each pair > >>>>>>of them has been emitted around bdrv_aligned_preadv(), and the new branch > >>>>>>doesn't do it anymore. So I don't see a reason to add events here. > >>>>>Yes, if you can assume that anyone who uses the debug events know > >>>>>exactly what the code looks like, adding the events here is pointless > >>>>>because TAIL, AFTER_TAIL and for the greatest part also AFTER_HEAD are > >>>>>essentially the same then. > >>>>> > >>>>>Having TAIL before the qiov change and AFTER_TAIL afterwards doesn't > >>>>>make any difference, they could (and should) be called immediately one > >>>>>after another if we wanted to keep the behaviour. > >>>>> > >>>>>I would agree that we should take a look at the test case and what it > >>>>>actually wants to achieve before we can decide whether AFTER_HEAD and > >>>>>TAIL/AFTER_TAIL would be the same (the former could trigger earlier if > >>>>>there are two requests and only one is unaligned at the tail). Maybe we > >>>>>even need to extend the test case now so that both paths (explicit read > >>>>>of the tail and the shortcut) are covered. > >>>>The part that actually blocks in 077 is > >>>> > >>>># Sequential RMW requests on the same physical sector > >>>> > >>>>its expecting all 4 events around the RMW cycle. > >>>> > >>>>However, it seems that also other parts of 077 would need an adjustment > >>>>and the output might differ depending on the alignment. So I guess we > >>>>have to emit the events if we don't want to recode the whole 077 and make > >>>>it aware of the alignment. > >>>Yes, but my point is that we may need to rework 077 anyway if we don't > >>>only want to make it pass again, but to cover all relevant paths, too. > >>>We got a new code path and it's unlikely that the existing tests covered > >>>both the old code path and the new one. > >>So you would postpone this patch until 077 is reworked? > >>I found this one a nice improvement and 077 might take some time. > >The problem with "we'll rework the tests later" is always that it > >doesn't happen if the patches for the functional parts and a workaround > >for the test case are merged. > > > >I don't think that making 077 cover both cases should be hard or take > >much time, it just needs to be done. If all the time for writing emails > >in this thread had been used to work on the test case, it would already > >be done. > > Understood. If you can give a hint how to get the value of the align > parameter into the test script I can try. Otherwise the test will fail > also if any block driver has an align value that is not equal to 512. The test case already uses blkdebug to enforce a specific align value (which is 4096 in this test case, not 512): echo "open -o driver=$IMGFMT,file.align=4k blkdebug::$TEST_IMG" Kevin