user_msqid_ds, user_semid_ds, user_shmid_ds
and
old_user_msqid_ds, old_user_semid_ds, old_user_shmid_ds
The former are used for the "new" IPC_STAT, IPC_SET, et. al, while the
latter are used for backwards compatibility (i.e. 16-bit UIDs).
I favored this approach because it made adding 32-bit UID support as
simple as possible. I've created functions such as
kernel2user_semid_ds(struct semid_ds *in, struct user_semid_ds *out)
kernel2user_ipc_perm(struct ipc_perm *in, struct user_ipc_perm *out)
kernel2old_user_ipc_perm(struct ipc_perm *in, struct old_user_ipc_perm *out)
rather than toss all the new code into the already hairy msgctl(),
semctl(), and shmctl() functions. This makes the 32-bit UID patch much
smaller and easier to manage.
If anyone is interested in looking over this patch, you can find it at:
http://www.engin.umich.edu/caen/systems/Linux/code/misc/2.3/19991130/linux-ipc.patch
The patch doesn't define the user_ipc_perm, user_msqid_ds, user_semid_ds,
and user_shmid_ds structures; these are defined on an
architecture-by-architecture basis for greatest flexibility. You can
examine the architectural patches at:
http://www.engin.umich.edu/caen/systems/Linux/code/misc/2.3/19991130/
linux-alpha.patch
linux-arm.patch
linux-i386.patch
linux-m68k.patch
linux-mips.patch
linux-ppc.patch
linux-sh.patch
linux-sparc.patch
to see the actual definitions.
Questions:
- does anyone think that this is a bad idea?
- should there be more pad space left in the user_ipc_perm,
user_msqid_ds, user_semid_ds, and user_shmid_ds structures for
future use? At present, I've left 2 machine words worth of pad
space in the msqid, semid, and shmid structures, and no extra
padding in user_ipc_perm.
Thanks,
Chris Wing
wingc@engin.umich.edu
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/