Re: + syscallsx86-implement-execveat-system-call.patch added to -mm tree

Re: + syscallsx86-implement-execveat-system-call.patch added to -mm tree

[Oleg Nesterov]

> @@ -1479,7 +1489,26 @@ static int do_execve_common(struct filen
>
>  	bprm->file = file;
> -	bprm->filename = bprm->interp = filename->name;
> +	if (fd == AT_FDCWD || filename->name[0] == '/') {
> +		bprm->filename = filename->name;
> +	} else {
> +		if (filename->name[0] == '\0')
> +			pathbuf = kasprintf(GFP_TEMPORARY, "/dev/fd/%d", fd);
> +		else
> +			pathbuf = kasprintf(GFP_TEMPORARY, "/dev/fd/%d/%s",
> +					    fd, filename->name);
> +		if (!pathbuf) {
> +			retval = -ENOMEM;
> +			goto out_unmark;
> +		}
> +		/* Record that a name derived from an O_CLOEXEC fd will be
> +		 * inaccessible after exec. Relies on having exclusive access to
> +		 * current->files (due to unshare_files above). */
> +		if (close_on_exec(fd, current->files->fdt))
> +			bprm->interp_flags |= BINPRM_FLAGS_PATH_INACCESSIBLE;
> +		bprm->filename = pathbuf;
+	}
+	bprm->interp = bprm->filename;

Not sure I understand this patch, will try to read later...

Just once question, don't we leak pathbuf if exec() succeeds?

OTOH, if it fails,

>  out_free:
>  	free_bprm(bprm);
> +	kfree(pathbuf);

Is it correct if we fail after bprm_change_interp() was called? It seems
that we can free interp == pathbuf twice?

Oleg.

Re: + syscallsx86-implement-execveat-system-call.patch added to -mm tree

[David Drysdale]

On Fri, Nov 14, 2014 at 12:11 AM, Oleg Nesterov <oleg@redhat.com> wrote:
>> @@ -1479,7 +1489,26 @@ static int do_execve_common(struct filen
>>
>>       bprm->file = file;
>> -     bprm->filename = bprm->interp = filename->name;
>> +     if (fd == AT_FDCWD || filename->name[0] == '/') {
>> +             bprm->filename = filename->name;
>> +     } else {
>> +             if (filename->name[0] == '\0')
>> +                     pathbuf = kasprintf(GFP_TEMPORARY, "/dev/fd/%d", fd);
>> +             else
>> +                     pathbuf = kasprintf(GFP_TEMPORARY, "/dev/fd/%d/%s",
>> +                                         fd, filename->name);
>> +             if (!pathbuf) {
>> +                     retval = -ENOMEM;
>> +                     goto out_unmark;
>> +             }
>> +             /* Record that a name derived from an O_CLOEXEC fd will be
>> +              * inaccessible after exec. Relies on having exclusive access to
>> +              * current->files (due to unshare_files above). */
>> +             if (close_on_exec(fd, current->files->fdt))
>> +                     bprm->interp_flags |= BINPRM_FLAGS_PATH_INACCESSIBLE;
>> +             bprm->filename = pathbuf;
> +       }
> +       bprm->interp = bprm->filename;
>
> Not sure I understand this patch, will try to read later...
>
> Just once question, don't we leak pathbuf if exec() succeeds?

Doh, yes.  I was sure I'd run this through kmemleak too, although
the evidence in front of me now clearly implies I didn't ...

> OTOH, if it fails,
>
>>  out_free:
>>       free_bprm(bprm);
>> +     kfree(pathbuf);
>
> Is it correct if we fail after bprm_change_interp() was called? It seems
> that we can free interp == pathbuf twice?

I think this is OK -- bprm_change_interp() changes bprm->interp to point to a
newly kstrdup'ed string, but leaves brpm->filename as pathbuf.  The former
then gets freed in free_bprm() (because it differs from filename == pathbuf),
and pathbuf is freed on the line afterwards.

> Oleg.
>