Re: [PATCH] x86/asm/entry/32: Update ENOSYS handling to match 64-bit logic

[Denys Vlasenko]

On 04/10/2015 05:44 PM, Andy Lutomirski wrote:
> On Fri, Apr 10, 2015 at 8:33 AM, Denys Vlasenko <dvlasenk@xxxxxxxxxx> wrote:
>> Sometime ago Andy changed 64-bit syscall logic so that pt_regs->ax is
>> initially set to -ENOSYS, and on exit from syscall, it is updated with
>> actual return value. This simplified logic there.
>>
>> This patch does the same for 32-bit syscall entry points.
>>
>> The check for %rax being too big is moved to be just before
>> the call insn which dispatches execution through syscall table.
>> There is no way to accidentally skip this check now by jumping
>> to a label after it. This allows to remove redundant checks
>> after e.g. ptrace.
>>
>> If %rax is too big, we just skip over the (call, write %rax to pt_regs->ax)
>> insn pair. pt_regs->ax remains set to -ENOSYS, and it gets returned
>> to userspace.
> 
> This looks okay, but I'll read it again later today.

Unfortunately, there is a mistake.

>> @@ -247,9 +248,7 @@ sysexit_from_sys_call:
>>         movl %ebx,%esi                  /* 2nd arg: 1st syscall arg */
>>         movl %eax,%edi                  /* 1st arg: syscall number */
>>         call __audit_syscall_entry
>> -       movl RAX(%rsp),%eax     /* reload syscall number */
>> -       cmpq $(IA32_NR_syscalls-1),%rax
>> -       ja ia32_badsys
>> +       movl ORIG_RAX(%rsp),%eax        /* reload syscall number */

this is correct, now syscall# is only in orig_ax, not in ax...

>>         call __audit_syscall_exit
>> -       movq RAX(%rsp),%rax     /* reload syscall return value */
>> +       movq ORIG_RAX(%rsp),%rax        /* reload syscall return value */

but here I'm wrong, this line should not be changed.
Will send v2 now.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/