[PATCH] increase buffer size in getmntent to 4096 bytes

[PATCH] increase buffer size in getmntent to 4096 bytes

[Rostislav Skudnov]

getmntent fails if the length of the mount entry exceeds the buffer
size. The new buffer size matches that of glibc.
---
Even after this patch, getmntent may still fail since the device and mount
paths may be up to PATH_MAX in length, and the mount options may
(theoretically) have unlimited length. The same would happen in glibc, though.

 src/misc/mntent.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/misc/mntent.c b/src/misc/mntent.c
index a16d652..5adb583 100644
--- a/src/misc/mntent.c
+++ b/src/misc/mntent.c
@@ -49,7 +49,7 @@ struct mntent *getmntent_r(FILE *f, struct mntent *mnt, char *linebuf, int bufle
 
 struct mntent *getmntent(FILE *f)
 {
-	static char linebuf[256];
+	static char linebuf[4096];
 	static struct mntent mnt;
 	return getmntent_r(f, &mnt, linebuf, sizeof linebuf);
 }
-- 
2.1.4

Re: [PATCH] increase buffer size in getmntent to 4096 bytes

[Natanael Copa]

On Mon, 17 Oct 2016 10:24:25 +0000
Rostislav Skudnov <rostislav@tuxera.com> wrote:

> getmntent fails if the length of the mount entry exceeds the buffer
> size. The new buffer size matches that of glibc.
> ---
> Even after this patch, getmntent may still fail since the device and mount
> paths may be up to PATH_MAX in length, and the mount options may
> (theoretically) have unlimited length. The same would happen in glibc, though.

This patch solve the same problem, but uses getline to malloc the buffer as needed:
http://www.openwall.com/lists/musl/2016/09/08/1


-nc

Re: [PATCH] increase buffer size in getmntent to 4096 bytes

[Rich Felker]

On Mon, Oct 17, 2016 at 02:48:17PM +0200, Natanael Copa wrote:
> On Mon, 17 Oct 2016 10:24:25 +0000
> Rostislav Skudnov <rostislav@tuxera.com> wrote:
> 
> > getmntent fails if the length of the mount entry exceeds the buffer
> > size. The new buffer size matches that of glibc.
> > ---
> > Even after this patch, getmntent may still fail since the device and mount
> > paths may be up to PATH_MAX in length, and the mount options may
> > (theoretically) have unlimited length. The same would happen in glibc, though.
> 
> This patch solve the same problem, but uses getline to malloc the buffer as needed:
> http://www.openwall.com/lists/musl/2016/09/08/1

Indeed. We can't go making every program use an additional page of
memory just for a rare program that uses getmntent, so dynamic
allocation it had to be.

Rich