This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: misc/tst-ttyname time outs
* Szabolcs Nagy:
> On 02/01/2019 18:51, Florian Weimer wrote:
>> * Carlos O'Donell:
>>> "info: posix_openpt: Failed with errno %d. Consider increasing limits."
>>
>> Yes, that's what I meant.
>>
>> I don't think we should start printing sysctls and mount options here
>> (or systemd-nspawn arguments).
>
> makes sense to me.
>
>> This ENOSPC comes from the max= mount option on the devpts instance, I
>> think. So it's different.
>
> thanks, this was not clear to me.
>
> EMFILE happens here when the slavename pts number > 1023,
> i don't know how those numbers are allocated and how to
> make them smaller.
>
> [pid 16749] newfstatat(AT_FDCWD, "/dev/pts/1852", 0xffffdb4ff040, 0) = -1 ENOENT (No such file or directory)
> [pid 16749] openat(AT_FDCWD, "/dev/ptmx", O_RDWR|O_NOCTTY|O_NONBLOCK) = 1022
> [pid 16749] newfstatat(AT_FDCWD, "/dev/pts/1852", 0xffffdb4ff040, 0) = -1 ENOENT (No such file or directory)
> [pid 16749] openat(AT_FDCWD, "/dev/ptmx", O_RDWR|O_NOCTTY|O_NONBLOCK) = 1023
> [pid 16749] newfstatat(AT_FDCWD, "/dev/pts/1852", 0xffffdb4ff040, 0) = -1 ENOENT (No such file or directory)
> [pid 16749] openat(AT_FDCWD, "/dev/ptmx", O_RDWR|O_NOCTTY|O_NONBLOCK) = -1 EMFILE (Too many open files)
> [pid 16749] newfstatat(AT_FDCWD, "/dev/pts/1852", 0xffffdb4ff040, 0) = -1 ENOENT (No such file or directory)
> [pid 16749] openat(AT_FDCWD, "/dev/ptmx", O_RDWR|O_NOCTTY|O_NONBLOCK) = -1 EMFILE (Too many open files)
> ...
>
> ENOSPC happens when i run multiple instances of this test
> (with --direct so they are in infinite loop) so pty limit
> explains it:
>
> $ sysctl kernel.pty
> kernel.pty.max = 4096
> kernel.pty.nr = 1864
> kernel.pty.reserve = 1024
>
> mount options:
> udev on /dev type devtmpfs (rw,nosuid,relatime,size=65955416k,nr_inodes=16488854,mode=755)
> devpts on /dev/pts type devpts (rw,nosuid,noexec,relatime,gid=5,mode=620,ptmxmode=000)
ENOSPC is tough to deal with here. I suspect it involves asynchronous
namespace deallocation (so the next run of the test starts before the
previous namespace is gone).
For EMFILE, please try the attached patch. It should help.
It also guards against the infinite loop.
Thanks,
Florian
Subject: [PATCH] Linux: Improve handling of resource limits in misc/tst-ttyname
An attempt to re-create a different PTY under the same name can fail
if the PTY has a very high number. Try to increase the file
descriptor limit in this case, and bail out if this still does not
allow the test to proceed.
2019-01-03 Florian Weimer <fweimer@redhat.com>
* sysdeps/unix/sysv/linux/tst-ttyname.c (adjust_file_limit): New
function.
(do_in_chroot_1): Call it.
(run_chroot_tests):
Improve error reporting in case it is not possible to create a
collision for the PTY name required by the test.
diff --git a/sysdeps/unix/sysv/linux/tst-ttyname.c b/sysdeps/unix/sysv/linux/tst-ttyname.c
index 5d2a54fbbf..7743699816 100644
--- a/sysdeps/unix/sysv/linux/tst-ttyname.c
+++ b/sysdeps/unix/sysv/linux/tst-ttyname.c
@@ -27,6 +27,7 @@
#include <sys/prctl.h>
#include <sys/stat.h>
#include <sys/wait.h>
+#include <sys/resource.h>
#include <unistd.h>
#include <support/check.h>
@@ -239,6 +240,32 @@ prepare (int argc, char **argv)
}
#define PREPARE prepare
+/* Adjust the file limit so that we have a chance to open PTY. */
+static void
+adjust_file_limit (const char *pty)
+{
+ int number = -1;
+ if (sscanf (pty, "/dev/pts/%d", &number) != 1 || number < 0)
+ FAIL_EXIT1 ("invalid PTY name: \"%s\"", pty);
+
+ /* Add a few additional descriptors to cover standard I/O streams
+ etc. */
+ rlim_t desired_limit = number + 10;
+
+ struct rlimit lim;
+ if (getrlimit (RLIMIT_NOFILE, &lim) != 0)
+ FAIL_EXIT1 ("getrlimit (RLIMIT_NOFILE): %m");
+ if (lim.rlim_cur < desired_limit)
+ {
+ printf ("info: adjusting RLIMIT_NOFILE from %llu to %llu\n",
+ (unsigned long long int) lim.rlim_cur,
+ (unsigned long long int) desired_limit);
+ lim.rlim_cur = desired_limit;
+ if (setrlimit (RLIMIT_NOFILE, &lim) != 0)
+ printf ("warning: setrlimit (RLIMIT_NOFILE) failed: %m\n");
+ }
+}
+
/* These chroot setup functions put the TTY at at "/console" (where it
won't be found by ttyname), and create "/dev/console" as an
ordinary file. This way, it's easier to write test-cases that
@@ -266,6 +293,7 @@ do_in_chroot_1 (int (*cb)(const char *, int))
if (strncmp (slavename, "/dev/pts/", 9) != 0)
FAIL_UNSUPPORTED ("slave pseudo-terminal is not under /dev/pts/: %s",
slavename);
+ adjust_file_limit (slavename);
int slave = xopen (slavename, O_RDWR, 0);
if (!doit (slave, "basic smoketest",
(struct result_r){.name=slavename, .ret=0, .err=0}))
@@ -332,6 +360,7 @@ do_in_chroot_2 (int (*cb)(const char *, int))
if (strncmp (slavename, "/dev/pts/", 9) != 0)
FAIL_UNSUPPORTED ("slave pseudo-terminal is not under /dev/pts/: %s",
slavename);
+ adjust_file_limit (slavename);
/* wait until in a new mount ns to open the slave */
/* enable `wait`ing on grandchildren */
@@ -445,10 +474,20 @@ run_chroot_tests (const char *slavename, int slave)
ok = false;
VERIFY (umount ("/dev/console") == 0);
- /* keep creating PTYs until we we get a name collision */
- while (stat (slavename, &st) < 0)
- posix_openpt (O_RDWR|O_NOCTTY|O_NONBLOCK);
- VERIFY (stat (slavename, &st) == 0);
+ /* Keep creating PTYs until we we get a name collision. */
+ while (true)
+ {
+ if (stat (slavename, &st) == 0)
+ break;
+ if (posix_openpt (O_RDWR|O_NOCTTY|O_NONBLOCK) < 0)
+ {
+ if (errno == ENOSPC || errno == EMFILE || errno == ENOSPC)
+ FAIL_UNSUPPORTED ("cannot re-create PTY \"%s\" in chroot: %m"
+ " (consider increasing limits)", slavename);
+ else
+ FAIL_EXIT1 ("cannot re-create PTY \"%s\" chroot: %m", slavename);
+ }
+ }
if (!doit (slave, "conflict, no match",
(struct result_r){.name=NULL, .ret=ENODEV, .err=ENODEV}))