From: "Rao, Lei" <lei.rao@intel.com>
To: "lizhijian@fujitsu.com" <lizhijian@fujitsu.com>,
"chen.zhang@intel.com" <chen.zhang@intel.com>,
"jasowang@redhat.com" <jasowang@redhat.com>
Cc: "qemu-devel@nongnu.org" <qemu-devel@nongnu.org>
Subject: Re: [PATCH] net/filter: Optimize filter_send to coroutine
Date: Fri, 24 Dec 2021 20:25:10 +0800 [thread overview]
Message-ID: <c385fe27-0721-144f-71dc-5437142c1e82@intel.com> (raw)
In-Reply-To: <b6233b02-21f0-be63-7326-c23ec3c0fe08@fujitsu.com>
On 12/24/2021 6:07 PM, lizhijian@fujitsu.com wrote:
>
>
> On 24/12/2021 10:37, Rao, Lei wrote:
>> This patch is to improve the logic of QEMU main thread sleep code in
>> qemu_chr_write_buffer() where it can be blocked and can't run other
>> coroutines during COLO IO stress test.
>>
>> Our approach is to put filter_send() in a coroutine. In this way,
>> filter_send() will call qemu_coroutine_yield() in qemu_co_sleep_ns(),
>> so that it can be scheduled out and QEMU main thread has opportunity to
>> run other tasks.
>>
>> Signed-off-by: Lei Rao <lei.rao@intel.com>
>> Signed-off-by: Zhang Chen <chen.zhang@intel.com>
>> ---
>> net/filter-mirror.c | 67 ++++++++++++++++++++++++++++++++++++---------
>> 1 file changed, 54 insertions(+), 13 deletions(-)
>>
>> diff --git a/net/filter-mirror.c b/net/filter-mirror.c
>> index f20240cc9f..1e9f8b6216 100644
>> --- a/net/filter-mirror.c
>> +++ b/net/filter-mirror.c
>> @@ -20,6 +20,7 @@
>> #include "chardev/char-fe.h"
>> #include "qemu/iov.h"
>> #include "qemu/sockets.h"
>> +#include "block/aio-wait.h"
>>
>> #define TYPE_FILTER_MIRROR "filter-mirror"
>> typedef struct MirrorState MirrorState;
>> @@ -42,20 +43,21 @@ struct MirrorState {
>> bool vnet_hdr;
>> };
>>
>> -static int filter_send(MirrorState *s,
>> - const struct iovec *iov,
>> - int iovcnt)
>> +typedef struct FilterSendCo {
>> + MirrorState *s;
>> + char *buf;
>> + ssize_t size;
>> + bool done;
>> + int ret;
>> +} FilterSendCo;
>> +
>> +static int _filter_send(MirrorState *s,
>> + char *buf,
>> + ssize_t size)
>> {
>> NetFilterState *nf = NETFILTER(s);
>> int ret = 0;
>> - ssize_t size = 0;
>> uint32_t len = 0;
>> - char *buf;
>> -
>> - size = iov_size(iov, iovcnt);
>> - if (!size) {
>> - return 0;
>> - }
>>
>> len = htonl(size);
>> ret = qemu_chr_fe_write_all(&s->chr_out, (uint8_t *)&len, sizeof(len));
>> @@ -80,10 +82,7 @@ static int filter_send(MirrorState *s,
>> }
>> }
>>
>> - buf = g_malloc(size);
>> - iov_to_buf(iov, iovcnt, 0, buf, size);
>> ret = qemu_chr_fe_write_all(&s->chr_out, (uint8_t *)buf, size);
>> - g_free(buf);
>> if (ret != size) {
>> goto err;
>> }
>> @@ -94,6 +93,48 @@ err:
>> return ret < 0 ? ret : -EIO;
>> }
>>
>> +static void coroutine_fn filter_send_co(void *opaque)
>> +{
>> + FilterSendCo *data = opaque;
>> +
>> + data->ret = _filter_send(data->s, data->buf, data->size);
>> + data->done = true;
>> + g_free(data->buf);
>> + aio_wait_kick();
>> +}
>> +
>> +static int filter_send(MirrorState *s,
>> + const struct iovec *iov,
>> + int iovcnt)
>> +{
>> + ssize_t size = iov_size(iov, iovcnt);
>> + char *buf = NULL;
>> +
>> + if (!size) {
>> + return 0;
>> + }
>> +
>> + buf = g_malloc(size);
>> + iov_to_buf(iov, iovcnt, 0, buf, size);
>> +
>> + FilterSendCo data = {
>> + .s = s,
>> + .size = size,
>> + .buf = buf,
>> + .ret = 0,
>> + };
>> +
>> + Coroutine *co = qemu_coroutine_create(filter_send_co, &data);
>
> BTW, does qemu/old gcc complaint such coding style ?
>
> int a;
> a = foo()
> int b = a;
There are a lot of codes of this style in QEMU.
It is written that we need at least GCC v7.4 to compile QEMU in the configure file.
So, I think it is no problem.
>
>
>
>> + qemu_coroutine_enter(co);
>> +
>> + while (!data.done) {
>> + aio_poll(qemu_get_aio_context(), true);
>> + }
>> +
>> + return data.ret;
>> +
> redundant newline
will be changed in V2.
Thanks,
Lei
>
> Otherwise,
> Reviewed-by: Li Zhijian <lizhijian@fujitsu.com>
>
>
>
>> +}
>> +
>> static void redirector_to_filter(NetFilterState *nf,
>> const uint8_t *buf,
>> int len)
prev parent reply other threads:[~2021-12-24 12:26 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-12-24 2:37 [PATCH] net/filter: Optimize filter_send to coroutine Rao, Lei
2021-12-24 10:07 ` lizhijian
2021-12-24 12:25 ` Rao, Lei [this message]
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=c385fe27-0721-144f-71dc-5437142c1e82@intel.com \
--to=lei.rao@intel.com \
--cc=chen.zhang@intel.com \
--cc=jasowang@redhat.com \
--cc=lizhijian@fujitsu.com \
--cc=qemu-devel@nongnu.org \
/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).