linux/semctl: fix SEM_STAT_ANY support [BZ #26637]

Message ID 20200920180529.GA1484@altlinux.org
State New
Headers show
Series
  • linux/semctl: fix SEM_STAT_ANY support [BZ #26637]
Related show

Commit Message

Dmitry V. Levin Sept. 20, 2020, 6:05 p.m.
Handle SEM_STAT_ANY the same way as SEM_STAT so that the buffer argument
of SEM_STAT_ANY is properly passed to the kernel and back.

* sysdeps/unix/sysv/linux/semctl.c (semun64_to_ksemun64,
semun_to_semun64, __semctl, __semctl_mode16, __old_semctl): Handle
SEM_STAT_ANY.
---
 sysdeps/unix/sysv/linux/semctl.c | 6 ++++++
 1 file changed, 6 insertions(+)

-- 
ldv

Comments

Adhemerval Zanella via Libc-alpha Sept. 25, 2020, 11:51 a.m. | #1
* Dmitry V. Levin:

> Handle SEM_STAT_ANY the same way as SEM_STAT so that the buffer argument

> of SEM_STAT_ANY is properly passed to the kernel and back.

>

> * sysdeps/unix/sysv/linux/semctl.c (semun64_to_ksemun64,

> semun_to_semun64, __semctl, __semctl_mode16, __old_semctl): Handle

> SEM_STAT_ANY.


As a stop-gap measure, this is okay, but we really need to reconsider
what we are doing here.  We really need to return EINVAL if we do not
recognize the cmd value.  It means patching glibc each time the system
call is enhanced, but I don't see a way around that.

Thanks,
Florian
-- 
Red Hat GmbH, https://de.redhat.com/ , Registered seat: Grasbrunn,
Commercial register: Amtsgericht Muenchen, HRB 153243,
Managing Directors: Charles Cachera, Brian Klemm, Laurie Krebs, Michael O'Neill
Adhemerval Zanella via Libc-alpha Sept. 25, 2020, 5:12 p.m. | #2
As a side note I am also working on this patch as well with BZ#26636
and BZ#26639
to add a regression test.

On Fri, Sep 25, 2020 at 8:51 AM Florian Weimer via Libc-alpha
<libc-alpha@sourceware.org> wrote:
>

> * Dmitry V. Levin:

>

> > Handle SEM_STAT_ANY the same way as SEM_STAT so that the buffer argument

> > of SEM_STAT_ANY is properly passed to the kernel and back.

> >

> > * sysdeps/unix/sysv/linux/semctl.c (semun64_to_ksemun64,

> > semun_to_semun64, __semctl, __semctl_mode16, __old_semctl): Handle

> > SEM_STAT_ANY.

>

> As a stop-gap measure, this is okay, but we really need to reconsider

> what we are doing here.  We really need to return EINVAL if we do not

> recognize the cmd value.  It means patching glibc each time the system

> call is enhanced, but I don't see a way around that.

>

> Thanks,

> Florian

> --

> Red Hat GmbH, https://de.redhat.com/ , Registered seat: Grasbrunn,

> Commercial register: Amtsgericht Muenchen, HRB 153243,

> Managing Directors: Charles Cachera, Brian Klemm, Laurie Krebs, Michael O'Neill

>

Patch

diff --git a/sysdeps/unix/sysv/linux/semctl.c b/sysdeps/unix/sysv/linux/semctl.c
index f131a26fc7..1cdabde8f2 100644
--- a/sysdeps/unix/sysv/linux/semctl.c
+++ b/sysdeps/unix/sysv/linux/semctl.c
@@ -102,6 +102,7 @@  semun64_to_ksemun64 (int cmd, union semun64 semun64,
       r.array = semun64.array;
       break;
     case SEM_STAT:
+    case SEM_STAT_ANY:
     case IPC_STAT:
     case IPC_SET:
       r.buf = buf;
@@ -150,6 +151,7 @@  __semctl64 (int semid, int semnum, int cmd, ...)
     case IPC_STAT:      /* arg.buf */
     case IPC_SET:
     case SEM_STAT:
+    case SEM_STAT_ANY:
     case IPC_INFO:      /* arg.__buf */
     case SEM_INFO:
       va_start (ap, cmd);
@@ -238,6 +240,7 @@  semun_to_semun64 (int cmd, union semun semun, struct __semid64_ds *semid64)
       r.array = semun.array;
       break;
     case SEM_STAT:
+    case SEM_STAT_ANY:
     case IPC_STAT:
     case IPC_SET:
       r.buf = semid64;
@@ -267,6 +270,7 @@  __semctl (int semid, int semnum, int cmd, ...)
     case IPC_STAT:      /* arg.buf */
     case IPC_SET:
     case SEM_STAT:
+    case SEM_STAT_ANY:
     case IPC_INFO:      /* arg.__buf */
     case SEM_INFO:
       va_start (ap, cmd);
@@ -321,6 +325,7 @@  __semctl_mode16 (int semid, int semnum, int cmd, ...)
     case IPC_STAT:      /* arg.buf */
     case IPC_SET:
     case SEM_STAT:
+    case SEM_STAT_ANY:
     case IPC_INFO:      /* arg.__buf */
     case SEM_INFO:
       va_start (ap, cmd);
@@ -354,6 +359,7 @@  __old_semctl (int semid, int semnum, int cmd, ...)
     case IPC_STAT:      /* arg.buf */
     case IPC_SET:
     case SEM_STAT:
+    case SEM_STAT_ANY:
     case IPC_INFO:      /* arg.__buf */
     case SEM_INFO:
       va_start (ap, cmd);