| Message ID | 4A8CAAE2.1030804@cn.fujitsu.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Thu, Aug 20, 2009 at 09:46:10AM +0800, Gui Jianfeng wrote: > Vivek Goyal wrote: > ... > > /* > > * Remember that we saw a request from this process, but > > @@ -1940,7 +2013,7 @@ void elv_ioq_request_add(struct request_queue *q, struct request *rq) > > * has other work pending, don't risk delaying until the > > * idle timer unplug to continue working. > > */ > > - if (elv_ioq_wait_request(ioq)) { > > + if (group_wait || elv_ioq_wait_request(ioq)) { > > Hi Vivek, > > I guess we need to clear ioq_wait_request flags if there are requests to > go in this ioq. Otherwise, once waitting request on ioq, it will go into > this path every time when a request is enqueued. > > Signed-off-by: Gui Jianfeng <guijianfeng@cn.fujitsu.com> Hi Gui, It does sound like that we need to clear ioq wait_request flag here. In fact upstream CFQ code is also not clearing this flag. Looking at the code, can't think why it should not be cleared here. Can you please also generate a patch for CFQ and post to lkml. Thanks Vivek > --- > block/elevator-fq.c | 3 +++ > 1 files changed, 3 insertions(+), 0 deletions(-) > > diff --git a/block/elevator-fq.c b/block/elevator-fq.c > index b3c387d..201543e 100644 > --- a/block/elevator-fq.c > +++ b/block/elevator-fq.c > @@ -2708,6 +2708,9 @@ void elv_ioq_request_add(struct request_queue *q, struct request *rq) > __blk_run_queue(q); > else > elv_mark_ioq_must_dispatch(ioq); > + > + if (elv_ioq_wait_request(ioq)) > + elv_clear_ioq_wait_request(ioq); > } > } else if (elv_should_preempt(q, ioq, rq)) { > /* > -- > 1.5.4.rc3 -- dm-devel mailing list dm-devel@redhat.com https://www.redhat.com/mailman/listinfo/dm-devel
Vivek Goyal wrote: > On Thu, Aug 20, 2009 at 09:46:10AM +0800, Gui Jianfeng wrote: >> Vivek Goyal wrote: >> ... >>> /* >>> * Remember that we saw a request from this process, but >>> @@ -1940,7 +2013,7 @@ void elv_ioq_request_add(struct request_queue *q, struct request *rq) >>> * has other work pending, don't risk delaying until the >>> * idle timer unplug to continue working. >>> */ >>> - if (elv_ioq_wait_request(ioq)) { >>> + if (group_wait || elv_ioq_wait_request(ioq)) { >> Hi Vivek, >> >> I guess we need to clear ioq_wait_request flags if there are requests to >> go in this ioq. Otherwise, once waitting request on ioq, it will go into >> this path every time when a request is enqueued. >> >> Signed-off-by: Gui Jianfeng <guijianfeng@cn.fujitsu.com> > > Hi Gui, > > It does sound like that we need to clear ioq wait_request flag here. In > fact upstream CFQ code is also not clearing this flag. Looking at the > code, can't think why it should not be cleared here. > > Can you please also generate a patch for CFQ and post to lkml. > Sure, will do.
diff --git a/block/elevator-fq.c b/block/elevator-fq.c index b3c387d..201543e 100644 --- a/block/elevator-fq.c +++ b/block/elevator-fq.c @@ -2708,6 +2708,9 @@ void elv_ioq_request_add(struct request_queue *q, struct request *rq) __blk_run_queue(q); else elv_mark_ioq_must_dispatch(ioq); + + if (elv_ioq_wait_request(ioq)) + elv_clear_ioq_wait_request(ioq); } } else if (elv_should_preempt(q, ioq, rq)) { /*