[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH v3 1/4] scripts: add a script to generate syscall_nr.h
From: |
Laurent Vivier |
Subject: |
Re: [PATCH v3 1/4] scripts: add a script to generate syscall_nr.h |
Date: |
Mon, 16 Mar 2020 21:20:24 +0100 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.5.0 |
Le 16/03/2020 à 20:27, Taylor Simpson a écrit :
>
>
>> Le 16/03/2020 à 17:21, Taylor Simpson a écrit :
>>>
>>>
>>>> -----Original Message-----
>>>> From: Laurent Vivier <address@hidden>
>>>> Sent: Monday, March 16, 2020 3:56 AM
>>>> To: address@hidden
>>>> Cc: Laurent Vivier <address@hidden>; Richard Henderson
>>>> <address@hidden>; Riku Voipio <address@hidden>; Taylor
>>>> Simpson <address@hidden>; Alistair Francis
>>>> <address@hidden>
>>>> Subject: [PATCH v3 1/4] scripts: add a script to generate syscall_nr.h
>>>>
>>>> This script is needed for targets based on asm-generic syscall numbers
>>>> generation
>>>>
>>>> +
>>>> +filter_defines()
>>>> +{
>>>> + grep -e "#define __NR_" -e "#define __NR3264"
>>>
>>> For Hexagon, we're still running a 4.9 kernel. When I tried out this
>>> script on
>> that code base, there are a handful of these
>>> #undef __NR_syscalls
>>> #define __NR_syscalls 291
>>> This works fine with normal C preprocessing, and the last one wins.
>> However, when the #undef's are filtered out, it lease to build errors from
>> multiple #define's of TARGET_NR_syscalls. AFAIK, qemu doesn't use this, so
>> it should be OK to filter out. So, I changed the above line to
>>> grep -e "#define __NR_" -e "#define __NR3264" | grep -v
>> "__NR_syscalls"
>>>
>>> If you're OK incorporating that now, great! Otherwise, I'll add it to the
>> Hexagon RFC patch series until we get to a newer kernel.
>>
>> Sorry, I've just sent the pull-request.
>>
>> Add it to your RFC, please.
>
> Sure.
>
>>
>> It's why this script is only an helper, so it means once the file is
>> generated it can be edited before being committed.
>
> Oh, I thought the goal was to have them all generated. That's why I
> suggested generating the "Do not modify" comment.
The "Do not modify" is a good idea: people will re-run the script
instead of editing directly the file, but then it can be adjusted.
Thanks,
Laurent
[PATCH v3 2/4] linux-user, aarch64: sync syscall numbers with kernel v5.5, Laurent Vivier, 2020/03/16
[PATCH v3 4/4] linux-user, openrisc: sync syscall numbers with kernel v5.5, Laurent Vivier, 2020/03/16
[PATCH v3 3/4] linux-user, nios2: sync syscall numbers with kernel v5.5, Laurent Vivier, 2020/03/16
Re: [PATCH v3 0/4] linux-user: generate syscall_nr.h from linux unistd.h, no-reply, 2020/03/16