From: Edward Makarov <makarov@kraftway.ru>
To: Andrew Rybchenko <arybchenko@solarflare.com>,
Vipul Ashri <vipul.ashri@oracle.com>,
"dev@dpdk.org" <dev@dpdk.org>
Cc: "chenbo.xia@intel.com" <chenbo.xia@intel.com>,
"zhihong.wang@intel.com" <zhihong.wang@intel.com>,
"maxime.coquelin@redhat.com" <maxime.coquelin@redhat.com>,
"stable@dpdk.org" <stable@dpdk.org>
Subject: Re: [dpdk-stable] [dpdk-dev] [PATCH] net/virtio: fix wrong variable assignment in helper macro
Date: Sat, 29 Aug 2020 22:17:54 +0000 [thread overview]
Message-ID: <f2538056-20ba-d957-2f12-904c7702047a@kraftway.ru> (raw)
In-Reply-To: <7a2d66d6-3961-5251-36d2-c702fe80379f@solarflare.com>
On 8/29/20 2:22 PM, Andrew Rybchenko wrote:
> On 8/14/20 12:21 PM, Vipul Ashri wrote:
>> Inside Macro ASSIGN_UNLESS_EQUAL(var, val), assignment to var is always
>> failing as assignment done using var_ having local scope only.
>> This leads to TX packets not going out and found broken due to cleanup
>> malfunctioning. This patch fixes the wrong variable assignment.
>>
>> Fixes: 57f90f894588 ("net/virtio: reuse packed ring functions")
>> Cc: stable@dpdk.org
>>
>> Signed-off-by: Vipul Ashri <vipul.ashri@oracle.com>
>> ---
>> drivers/net/virtio/virtqueue.h | 6 ++----
>> 1 file changed, 2 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/net/virtio/virtqueue.h b/drivers/net/virtio/virtqueue.h
>> index 105a9c00c..20c95471e 100644
>> --- a/drivers/net/virtio/virtqueue.h
>> +++ b/drivers/net/virtio/virtqueue.h
>> @@ -607,10 +607,8 @@ virtqueue_notify(struct virtqueue *vq)
>>
>> /* avoid write operation when necessary, to lessen cache issues */
>> #define ASSIGN_UNLESS_EQUAL(var, val) do { \
>> - typeof(var) var_ = (var); \
>> - typeof(val) val_ = (val); \
>> - if ((var_) != (val_)) \
>> - (var_) = (val_); \
>> + if ((var) != (val)) \
>> + (var) = (val); \
>
> Good catch. As I understand the old code tries to avoid
> processing of var and val expressions twice. It looks it
> could be kept for val at least. Just keep if condition as in
> old code and fix the last line above:
> (var) = val_;
> Since var_ and val_are local variables there is no necessity
> to enclose it in parenthesis (but does not harm if done).
> var_ may be really removed since since resulting code will
> use it only once.
I think there is a solution to avoid multiple evaluations of parameters:
// var is definitely an lvalue, so its address can definitely be taken
#define ASSIGN_UNLESS_EQUAL(var, val) do { \
typeof(var) *const var_ = &(var); \
typeof(val) const val_ = (val); \
if (*var_ != val_) \
*var_ = val_; \
} while (0)
The solution relies on the compiler optimizer. Here is the comparison of
what the variants are compiled into:
https://godbolt.org/z/nnvq5q
next prev parent reply other threads:[~2020-08-29 22:17 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <20200814092138.1692-1-vipul.ashri@oracle.com>
2020-08-29 11:22 ` Andrew Rybchenko
2020-08-29 22:17 ` Edward Makarov [this message]
2020-09-18 10:46 ` Vipul Ashri
2020-09-18 10:53 ` Maxime Coquelin
2020-09-18 10:41 ` Vipul Ashri
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=f2538056-20ba-d957-2f12-904c7702047a@kraftway.ru \
--to=makarov@kraftway.ru \
--cc=arybchenko@solarflare.com \
--cc=chenbo.xia@intel.com \
--cc=dev@dpdk.org \
--cc=maxime.coquelin@redhat.com \
--cc=stable@dpdk.org \
--cc=vipul.ashri@oracle.com \
--cc=zhihong.wang@intel.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).