Re: [PATCH v6 4/6] selftests/mm/vm_util: robust write_file()

[Chunyu Hu]

On Sun, Mar 29, 2026 at 05:27:13PM +0800, Chunyu Hu wrote:
> On Fri, Mar 27, 2026 at 03:32:01PM +0300, Mike Rapoport wrote:
> > On Tue, Mar 24, 2026 at 09:33:14AM +0800, Chunyu Hu wrote:
> > > Add two more checks for buflen and numwritten. The buflen should be at
> > > least one, otherwise the 'buflen - 1' could underflow and cause trouble.
> > > The numwritten should be equal to 'buflen - 1'. The test will exit if
> > > any of these conditions aren't met.
> > > 
> > > Additionally, add more print information when a write failure occurs or
> > > a truncated write happens, providing clearer diagnostics.
> > > 
> > > Signed-off-by: Chunyu Hu <chuhu@xxxxxxxxxx>
> > > ---
> > > Changes in v6:
> > >   - save/restore errno before and after the close(), so the log could show
> > >     the correct error info on failure. Suggested by AI.
> > > Chagnes in v5:
> > >   - new patch for making improve on write_file. Add more safety checks and
> > >     diagnostics info in log
> > > ---
> > >  tools/testing/selftests/mm/vm_util.c | 12 ++++++++++--
> > >  1 file changed, 10 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/tools/testing/selftests/mm/vm_util.c b/tools/testing/selftests/mm/vm_util.c
> > > index ad96d19d1b85..572ccb99de8e 100644
> > > --- a/tools/testing/selftests/mm/vm_util.c
> > > +++ b/tools/testing/selftests/mm/vm_util.c
> > > @@ -767,15 +767,23 @@ int unpoison_memory(unsigned long pfn)
> > >  
> > >  void write_file(const char *path, const char *buf, size_t buflen)
> > >  {
> > > -	int fd;
> > > +	int fd, saved_errno;
> > >  	ssize_t numwritten;
> > > +	if (buflen < 1)
> > > +		ksft_exit_fail_msg("Incorrect buffer len: %zu\n", buflen);
> > >  
> > >  	fd = open(path, O_WRONLY);
> > >  	if (fd == -1)
> > >  		ksft_exit_fail_msg("%s open failed: %s\n", path, strerror(errno));
> > 
> > We have ksft_exit_fail_perror(), fits great here
> 
> the ksft_exit_fail_perror() don't accept the fmt string, it only accept
> a single string as parameter.
> 
> ```
> static inline __noreturn void ksft_exit_fail_perror(const char *msg)
> 
> ```
> 
> But we can change it also support va_list arg like 
> ksft_exit_fail_msg(msg, ...) does:
> 
> static inline __noreturn void ksft_exit_fail_perror(msg, ...)
> 
> I can have a try in next version.

Tried in v7.

In v7, I tried to change ksft_exit_fail_perror to support va_list args,
and I used vasrpintf() to print the message to the buf and then pass it
to ksft_exit_fail_msg(). But it requires _GNU_SOURCE can cause a build
issue in mm-unstable tree from the kernel test robot report. 

The vasprintf, the static buf[] (will trucate long message), or make
a copy of body of the ksft_exit_fail_msg and append a perror string,
none of these ways is clean, using ksft_exit_fail_msg may be still
a better way.

What do you think, Mike?

> 
> > 
> > >  
> > >  	numwritten = write(fd, buf, buflen - 1);
> > > +	saved_errno = errno;
> > >  	close(fd);
> > > +	errno = saved_errno;
> > >  	if (numwritten < 1)
> > > -		ksft_exit_fail_msg("Write failed\n");
> > > +		ksft_exit_fail_msg("%s write(%s) failed: %s\n", path, buf,
> > > +				strerror(errno));
> > 
> > and here.
> 
> Same answer with above coment.
> 
> > 
> > > +	if (numwritten != buflen - 1)
> > > +		ksft_exit_fail_msg("%s write(%s) is truncated, expected %zu bytes, got %zd bytes\n",
> > > +				path, buf, buflen - 1, numwritten);
> > >  }
> > > -- 
> > > 2.53.0
> > > 
> > 
> > -- 
> > Sincerely yours,
> > Mike.
> > 
>