2017-07-08 09:56:49

by Markus Trippelsdorf

Subject: commit 67d7ddded32 (waitid(2): leave copyout of siginfo to syscall itself) breaks glibc posix/tst-waitid

Since:
commit 67d7ddded322db99f451a7959d56ed6c70a6c4aa
Author: Al Viro <[email protected]>
Date: Sun May 14 20:53:13 2017 -0400

waitid(2): leave copyout of siginfo to syscall itself

the glibc posix/tst-waitid.c testcase fails:

markus@x4 glibc-build % ./posix/tst-waitid
waitid WNOHANG on stopped status 0

--
Markus

2017-07-08 13:12:21

by Al Viro

Subject: Re: commit 67d7ddded32 (waitid(2): leave copyout of siginfo to syscall itself) breaks glibc posix/tst-waitid

On Sat, Jul 08, 2017 at 11:56:44AM +0200, Markus Trippelsdorf wrote:
> Since:
> commit 67d7ddded322db99f451a7959d56ed6c70a6c4aa
> Author: Al Viro <[email protected]>
> Date: Sun May 14 20:53:13 2017 -0400
>
> waitid(2): leave copyout of siginfo to syscall itself
>
> the glibc posix/tst-waitid.c testcase fails:
>
> markus@x4 glibc-build % ./posix/tst-waitid
> waitid WNOHANG on stopped status 0

I think I see what's going on... We lose the distinction between "got
a PID" and "nothing, but that's not an error" a bit too early waitid()
case. Could you see if there are any problems left with the following
applied?

diff --git a/kernel/exit.c b/kernel/exit.c
index 462fc25eec6e..aa5d5e4de804 100644
--- a/kernel/exit.c
+++ b/kernel/exit.c
@@ -1604,9 +1604,6 @@ static long kernel_waitid(int which, pid_t upid, struct waitid_info *infop,
wo.wo_rusage = ru;
ret = do_wait(&wo);

- if (ret > 0)
- ret = 0;
-
put_pid(pid);
return ret;
}
@@ -1617,6 +1614,11 @@ SYSCALL_DEFINE5(waitid, int, which, pid_t, upid, struct siginfo __user *,
struct rusage r;
struct waitid_info info = {.status = 0};
long err = kernel_waitid(which, upid, &info, options, ru ? &r : NULL);
+ int signo = 0;
+ if (err > 0) {
+ signo = SIGCHLD;
+ err = 0;
+ }

if (!err) {
if (ru && copy_to_user(ru, &r, sizeof(struct rusage)))
@@ -1626,7 +1628,7 @@ SYSCALL_DEFINE5(waitid, int, which, pid_t, upid, struct siginfo __user *,
return err;

user_access_begin();
- unsafe_put_user(err ? 0 : SIGCHLD, &infop->si_signo, Efault);
+ unsafe_put_user(signo, &infop->si_signo, Efault);
unsafe_put_user(0, &infop->si_errno, Efault);
unsafe_put_user((short)info.cause, &infop->si_code, Efault);
unsafe_put_user(info.pid, &infop->si_pid, Efault);
@@ -1728,6 +1730,11 @@ COMPAT_SYSCALL_DEFINE5(waitid,
struct rusage ru;
struct waitid_info info = {.status = 0};
long err = kernel_waitid(which, pid, &info, options, uru ? &ru : NULL);
+ int signo = 0;
+ if (err > 0) {
+ signo = SIGCHLD;
+ err = 0;
+ }

if (!err && uru) {
/* kernel_waitid() overwrites everything in ru */
@@ -1743,7 +1750,7 @@ COMPAT_SYSCALL_DEFINE5(waitid,
return err;

user_access_begin();
- unsafe_put_user(err ? 0 : SIGCHLD, &infop->si_signo, Efault);
+ unsafe_put_user(signo, &infop->si_signo, Efault);
unsafe_put_user(0, &infop->si_errno, Efault);
unsafe_put_user((short)info.cause, &infop->si_code, Efault);
unsafe_put_user(info.pid, &infop->si_pid, Efault);

2017-07-08 13:53:20

by Markus Trippelsdorf

Subject: Re: commit 67d7ddded32 (waitid(2): leave copyout of siginfo to syscall itself) breaks glibc posix/tst-waitid

On 2017.07.08 at 14:12 +0100, Al Viro wrote:
> On Sat, Jul 08, 2017 at 11:56:44AM +0200, Markus Trippelsdorf wrote:
> > Since:
> > commit 67d7ddded322db99f451a7959d56ed6c70a6c4aa
> > Author: Al Viro <[email protected]>
> > Date: Sun May 14 20:53:13 2017 -0400
> >
> > waitid(2): leave copyout of siginfo to syscall itself
> >
> > the glibc posix/tst-waitid.c testcase fails:
> >
> > markus@x4 glibc-build % ./posix/tst-waitid
> > waitid WNOHANG on stopped status 0
>
> I think I see what's going on... We lose the distinction between "got
> a PID" and "nothing, but that's not an error" a bit too early waitid()
> case. Could you see if there are any problems left with the following
> applied?

No, everything looks nice now. Thanks.

--
Markus