Hi,
I'm not sure if this really tests the branch where the issue is, but I
am also not able to come up with a simple test myself (since it
basically relies on predicting the random instantiation of the
template).
But this patch should avoid opening the file, if it already exists.
Cheers,
Sven
On Tue, 2024-02-06 at 15:13 +0100, Marius Melzer wrote:
> Hi,
>
> On 31.01.24 11:08, Waldemar Brodkorb wrote:
> > do you have a small C test case which might show the failure?
>
> I unfortunately currently don't have a proper setup to test this, but
> I
> would expect that this simple loop would deplete all available file
> descriptors so that at some point, one can't call "open" anymore:
>
> #include <stdlib.h>
>
> do {
> puts("Still fds available\n");
> } while(!strcmp(mktemp("test"),""))
>
> if(open("somefile.txt") < 0) {
> puts("fds are depleted. This should not have happened.\n");
> }
>
> Best regards,
> Marius
>
--
Sven Linker
Tel.: +49 351 41883243
Kernkonzept GmbH at Dresden, Germany,
HRB 31129, CEO Dr.-Ing. Michael Hohmuth
diff --git a/libc/misc/internals/tempname.c b/libc/misc/internals/tempname.c
index d3a8ccbd8..f9b714a68 100644
--- a/libc/misc/internals/tempname.c
+++ b/libc/misc/internals/tempname.c
@@ -218,7 +218,8 @@ int attribute_hidden __gen_tempname (char *tmpl, int kind, int flags,
/* Give up now. */
return -1;
} else
- fd = 0;
+ /* File already exists, so return with non-zero value */
+ return -1;
}
case __GT_FILE:
fd = open (tmpl, O_RDWR | O_CREAT | O_EXCL | flags, mode);
_______________________________________________
devel mailing list -- devel@uclibc-ng.org
To unsubscribe send an email to devel-le...@uclibc-ng.org
