On 7/26/26 15:17, Helge Deller wrote:
On 7/25/26 00:26, Pierrick Bouvier wrote:
On 7/24/2026 2:42 PM, Helge Deller wrote:
On 7/24/26 18:48, Pierrick Bouvier wrote:
On 7/24/2026 9:01 AM, no92 via qemu development wrote:
The members are declared as __kernel_pid_t in Linux UAPI headers.
Analogous members in struct target_shmid_ds (shm_[cl]pid) are also
declared as abi_int.
Fixes: 1c54ff97bbde ("linux-user: fix and cleanup IPCOP_msg* ipc
calls handling")
Signed-off-by: no92 <[email protected]>
---
linux-user/syscall.c | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/linux-user/syscall.c b/linux-user/syscall.c
index 3da5530d42..616d543c7f 100644
--- a/linux-user/syscall.c
+++ b/linux-user/syscall.c
@@ -4234,8 +4234,8 @@ struct target_msqid_ds
abi_ulong __msg_cbytes;
abi_ulong msg_qnum;
abi_ulong msg_qbytes;
- abi_ulong msg_lspid;
- abi_ulong msg_lrpid;
+ abi_int msg_lspid;
+ abi_int msg_lrpid;
...
The change looks correct in itself.
No.
abi_ulong is wrong, but abi_int isn't much better either if you look at
the kernel sources:
I wrongly assumed the definition we had on bsd-user side was correct,
but didn't really think about checking it in linux sources, sorry.
I need to correct myself :-(
abi_int *is* much better than anything else.
Indirectly all current builds use glibc which switches to __msqid64_ds
instead of the old msqid_ds struct, and in there __pid_t is used, which
seems to be defined to int for all targets.
So, leo's patch is correct and I'll pull it into my linux-user next git tree
as-is.
Thanks!
Helge
