Re: [PATCH 0/4] linux-user: Fix some issues in termbits.h files

From: Laurent Vivier <laurent@vivier.eu>
To: Aleksandar Markovic <aleksandar.markovic@rt-rk.com>,
	qemu-devel@nongnu.org
Cc: amarkovic@wavecomp.com
Subject: Re: [PATCH 0/4] linux-user: Fix some issues in termbits.h files
Date: Tue, 28 Jan 2020 19:19:20 +0100	[thread overview]
Message-ID: <994d96ac-ce43-9ff5-7212-eda6e12264fb@vivier.eu> (raw)
In-Reply-To: <1579227117-6310-1-git-send-email-aleksandar.markovic@rt-rk.com>

Le 17/01/2020 à 03:11, Aleksandar Markovic a écrit :
> From: Aleksandar Markovic <amarkovic@wavecomp.com>
> 
> This series is a spin-off of v5 of earlier series "linux-user: Misc
> patches for 5.0", that became too large to manage. I will submit the
> rest of that large series separately.
> 
> Files linux-user/<target>/termbits.h seem to be in a very bad shape:
> unsynchronized with kernel, containing wrong elements expressed in
> terms of host instead of target, many being updated wrt kernel
> content at various times, and on top of that all contain visually
> very ugly combinations of tabs and spaces.
> 
> This series attempts to fix great majority of issues in termbits.
> 
> Alpha's and mips' termbits.h were in the worst shape, missing large
> bits and pieces, and for them as complete as possible synchronization
> with kernel code is done - this constitutes the first two patches.
> 
> Xtensa's termbits.h contained the most elements wrongly expressed in
> terms of host instead of target, and that is the reason the changes
> in this file are placed in a separate, third, patch. Previous "R-B"
> given by Max Filippov was transferred to this patch only.
> 
> The fourth patch fixes remaining elements wrongly expressed in
> terms of host instead of target.
> 
> As an additional note, structures "serial_iso7816" and "serial_rs485"
> (at times mentioned as the third argument of certain ioctls) are
> platform-independant in kernel, and do not need "target_" variant
> in QEMU. Also, structure "winsize" (also appearing as the third
> ioctl's argument at times) is defined at multiple places in kernel
> (for several architectures) in kernel, but all such definitions are
> identical, and, therefore, it also does not need "target_" variant
> in QEMU.
> 
> A checkpatch warning related to "#if 0" in patch 2 is benign, and
> should be ignored.
> 
> Aleksandar Markovic (4):
>   linux-user: alpha: Synchronize termbits.h with kernel
>   linux-user: mips: Synchronize termbits.h with kernel
>   linux-user: xtensa: Fix some constants in termbits.h
>   linux-user: Fix some constants in remaining termbits.h
> 
>  linux-user/aarch64/termbits.h    |   4 +-
>  linux-user/alpha/termbits.h      |  82 ++++++++++++++--
>  linux-user/arm/termbits.h        |   4 +-
>  linux-user/cris/termbits.h       |   4 +-
>  linux-user/hppa/termbits.h       |   4 +-
>  linux-user/i386/termbits.h       |   4 +-
>  linux-user/m68k/termbits.h       |   4 +-
>  linux-user/microblaze/termbits.h |   4 +-
>  linux-user/mips/termbits.h       | 140 ++++++++++++++++----------
>  linux-user/nios2/termbits.h      |   4 +-
>  linux-user/openrisc/termbits.h   |  14 +--
>  linux-user/ppc/termbits.h        |   4 +-
>  linux-user/riscv/termbits.h      |   4 +-
>  linux-user/s390x/termbits.h      |  26 ++---
>  linux-user/sh4/termbits.h        |   4 +-
>  linux-user/sparc/termbits.h      |   4 +-
>  linux-user/sparc64/termbits.h    |   4 +-
>  linux-user/tilegx/termbits.h     |  12 ++-
>  linux-user/x86_64/termbits.h     |  26 +++--
>  linux-user/xtensa/termbits.h     | 207 +++++++++++++++++++++------------------
>  20 files changed, 353 insertions(+), 206 deletions(-)
> 

I think we should first introduce a linux-user/generic/termbits.h as we
have an asm-generic/termbits.h in the kernel and use it with all the
targets except alpha, mips, hppa, sparc and xtensa.

I think this linux-user/generic/termbits.h could be copied from
linux-user/openrisc/termbits.h (without the ioctl definitions)

Then you could update the remaining ones.

Thanks,
Laurent