* [9front] aux/wacom: possible race
@ 2021-12-06 2:50 qwx
2021-12-06 3:08 ` Stanley Lieber
2021-12-14 23:55 ` qwx
0 siblings, 2 replies; 4+ messages in thread
From: qwx @ 2021-12-06 2:50 UTC (permalink / raw)
To: 9front
Hello,
I've been playing with aux/wacom and aux/tablet on x61t.
aux/tablet often crashes with a `no concurrent reads, please'
error mid-drawing. From what I can tell both of these
programs are early 9front additions. Has anyone else seen
these errors?
I think that the issue is that nothing prevents the .req
field of each Reader from being modified at the same time
as a read(2) comes in, so at some point a new read could
check the field before sendout() can clear it. The patch
below seems to work.
What do you think?
Thanks,
qwx
diff 69249e8313fb8853c5c162dd16b927099e25bb15 uncommitted
--- a//sys/src/cmd/aux/wacom.c
+++ b//sys/src/cmd/aux/wacom.c
@@ -303,11 +303,13 @@
reply(req, m);
msgdecref(m);
} else {
+ lock(&readers);
if(r->req) {
respond(req, "no concurrent reads, please");
} else {
r->req = req;
}
+ unlock(&readers);
}
}
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [9front] aux/wacom: possible race
2021-12-06 2:50 [9front] aux/wacom: possible race qwx
@ 2021-12-06 3:08 ` Stanley Lieber
2021-12-14 23:55 ` qwx
1 sibling, 0 replies; 4+ messages in thread
From: Stanley Lieber @ 2021-12-06 3:08 UTC (permalink / raw)
To: 9front
On December 6, 2021 2:50:19 AM UTC, qwx@sciops.net wrote:
>Hello,
>
>I've been playing with aux/wacom and aux/tablet on x61t.
>aux/tablet often crashes with a `no concurrent reads, please'
>error mid-drawing. From what I can tell both of these
>programs are early 9front additions. Has anyone else seen
>these errors?
>
>I think that the issue is that nothing prevents the .req
>field of each Reader from being modified at the same time
>as a read(2) comes in, so at some point a new read could
>check the field before sendout() can clear it. The patch
>below seems to work.
>
>What do you think?
>
>Thanks,
>qwx
>
>
>diff 69249e8313fb8853c5c162dd16b927099e25bb15 uncommitted
>--- a//sys/src/cmd/aux/wacom.c
>+++ b//sys/src/cmd/aux/wacom.c
>@@ -303,11 +303,13 @@
> reply(req, m);
> msgdecref(m);
> } else {
>+ lock(&readers);
> if(r->req) {
> respond(req, "no concurrent reads, please");
> } else {
> r->req = req;
> }
>+ unlock(&readers);
> }
> }
>
>
no clue as to the cause, but i do occasionally see this on both x60t and x61t.
sl
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [9front] aux/wacom: possible race
2021-12-06 2:50 [9front] aux/wacom: possible race qwx
2021-12-06 3:08 ` Stanley Lieber
@ 2021-12-14 23:55 ` qwx
2021-12-15 0:00 ` Stanley Lieber
1 sibling, 1 reply; 4+ messages in thread
From: qwx @ 2021-12-14 23:55 UTC (permalink / raw)
To: 9front
On Mon Dec 6 02:57:05 +0100 2021, qwx@sciops.net wrote:
> Hello,
>
> I've been playing with aux/wacom and aux/tablet on x61t.
> aux/tablet often crashes with a `no concurrent reads, please'
> error mid-drawing. From what I can tell both of these
> programs are early 9front additions. Has anyone else seen
> these errors?
>
> I think that the issue is that nothing prevents the .req
> field of each Reader from being modified at the same time
> as a read(2) comes in, so at some point a new read could
> check the field before sendout() can clear it. The patch
> below seems to work.
>
> What do you think?
>
> Thanks,
> qwx
After testing, the errors no longer occur. I believe
that this is the right fix, and have pushed it.
Cheers!
qwx
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2021-12-15 0:08 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-12-06 2:50 [9front] aux/wacom: possible race qwx
2021-12-06 3:08 ` Stanley Lieber
2021-12-14 23:55 ` qwx
2021-12-15 0:00 ` Stanley Lieber
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).