Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Zhangjin Wu <falcon@tinylab.org>
To:     thomas@t-8ch.de, arnd@arndb.de, w@1wt.eu
Cc:     falcon@tinylab.org, linux-kernel@vger.kernel.org,
        linux-kselftest@vger.kernel.org, linux-riscv@lists.infradead.org
Subject: Re: [PATCH v2 04/13] tools/nolibc: add missing nanoseconds support for __NR_statx
Date:   Tue, 30 May 2023 13:21:32 +0800
Message-Id: <20230530052132.364685-1-falcon@tinylab.org>
In-Reply-To: <3a3edd48-1ace-4c89-89e8-9c594dd1b3c9@t-8ch.de>
References: <3a3edd48-1ace-4c89-89e8-9c594dd1b3c9@t-8ch.de>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Feedback-ID: bizesmtp:tinylab.org:qybglogicsvrsz:qybglogicsvrsz3a-3
Precedence: bulk

Thomas, Arnd, Willy

> On 2023-05-30 03:50:34+0800, Zhangjin Wu wrote:
> > Commit a89c937d781a ("tools/nolibc: support nanoseconds in stat()")
> > added nanoseconds for stat() but missed the statx case, this adds it.
> 
> Welp, I should have thought of that.
> At least the testcase seems to have been useful.
>

yeah, your testcase telled me this issue.

> Thanks for the fix!
> 
> > The stx_atime, stx_mtime, stx_ctime are in type of 'struct
> > statx_timestamp', which is incompatible with 'struct timespec', should
> > convert explicitly.
> > 
> >     /* include/uapi/linux/stat.h */
> > 
> >     struct statx_timestamp {
> >     	__s64	tv_sec;
> >     	__u32	tv_nsec;
> >     	__s32	__reserved;
> >     };
> > 
> >     /* include/uapi/linux/time_types.h */
> >     struct __kernel_timespec {
> >     	__kernel_time64_t       tv_sec;                 /* seconds */
> >     	long long               tv_nsec;                /* nanoseconds */
> >     };
> > 
> >     /* tools/include/nolibc/types.h */
> >     #define timespec __kernel_timespec
> > 
> > Without this patch, the stat_timestamps test case would fail on rv32.
> > 
> > Fixes: a89c937d781a ("tools/nolibc: support nanoseconds in stat()")
> > Signed-off-by: Zhangjin Wu <falcon@tinylab.org>
> > ---
> >  tools/include/nolibc/sys.h | 6 +++---
> >  1 file changed, 3 insertions(+), 3 deletions(-)
> > 
> > diff --git a/tools/include/nolibc/sys.h b/tools/include/nolibc/sys.h
> > index 154194056962..98cfa2f6d021 100644
> > --- a/tools/include/nolibc/sys.h
> > +++ b/tools/include/nolibc/sys.h
> > @@ -1175,9 +1175,9 @@ int sys_stat(const char *path, struct stat *buf)
> >  	buf->st_size    = statx.stx_size;
> >  	buf->st_blksize = statx.stx_blksize;
> >  	buf->st_blocks  = statx.stx_blocks;
> > -	buf->st_atime   = statx.stx_atime.tv_sec;
> > -	buf->st_mtime   = statx.stx_mtime.tv_sec;
> > -	buf->st_ctime   = statx.stx_ctime.tv_sec;
> > +	buf->st_atim    = (struct timespec){ .tv_sec = statx.stx_atime.tv_sec, .tv_nsec = statx.stx_atime.tv_nsec };
> > +	buf->st_mtim    = (struct timespec){ .tv_sec = statx.stx_mtime.tv_sec, .tv_nsec = statx.stx_mtime.tv_nsec };
> > +	buf->st_ctim    = (struct timespec){ .tv_sec = statx.stx_ctime.tv_sec, .tv_nsec = statx.stx_ctime.tv_nsec };
> 
> I would prefer to split the compound assignment into two single
> assignments, though.
> 
> buf->st_ctim.tv_sec = statx.stx_ctime.tv_sec;
> buf->st_ctim.tv_nsec = statx.stx_ctime.tv_nsec;
>

Ok, will update it in the v3 revision.

And further, what about removing the other !statx parts
(__NR_newfstatat, __NR_stat)? just like we are doing for the other 64bit
syscalls (llseek and time65).

Best regards,
Zhangjin

> >  	return ret;
> >  }
> >  #else