47626 - Purify on Solaris reports UMR and UMC in nsLocalFileUnix

Status: RESOLVED FIXED

(Core :: XPCOM, defect, P3)

Description

[Mark G. Adams]

Purify reports that the resolved_path local var in Normalize()
[nsLocalFileUnix.cpp 482] results in a UMR when realpath() is called. In
addition, a whole host of UMR/UMC's are returned from line 489. The following
patch fixes this.

I think that this is just a case of Purify being over-cautious; however, this
fix will at least clear out the noise.

------------------------------------------------------------------------
Index: nsLocalFileUnix.cpp
===================================================================
RCS file: /cvsroot/mozilla/xpcom/io/nsLocalFileUnix.cpp,v
retrieving revision 1.34
diff -u -r1.34 nsLocalFileUnix.cpp
--- nsLocalFileUnix.cpp 2000/08/03 22:17:43     1.34
+++ nsLocalFileUnix.cpp 2000/08/04 18:36:38
@@ -468,7 +468,7 @@
 NS_IMETHODIMP
 nsLocalFile::Normalize()
 {
-    char  resolved_path[PATH_MAX];
+    char  resolved_path[PATH_MAX] = "";
     char *resolved_path_ptr = NULL;
     CHECK_mPath();
 #ifdef XP_BEOS

Comment 1

[Gervase Markham [:gerv]]

Confirming for consideration.

Gerv

Comment 2

[Ray Whitmer]

I will worry about this after beta3.

Comment 3

[leger]

Updating QA Contact

Comment 4

[Margaret Chan]

Hi Ray:

I am in the process of learning how the checkin process works.  Would you
mind if I take this one from you since this seems to be simple enough
to test out.  Would like to get your approval to checkin later on.

Thanks, Margaret

Comment 5

[Warren Harris]

Edward: Welcome to xpcom!

Comment 6

[Warren Harris]

Once again... attempting to reassign from Ray to Edward.

Comment 7

[Tomi Leppikangas]

I see this with sun's workshop too. There isn't real bug, memory access check
just misses that realpath() somehow, maybe it does some woodoo magic in memory.

Mark's patch works ok and workshop stop whining. 

Comment 8

[Edward Kandrot]

I do not have a way to verify this patch.  Anyone who can take this?  Thanks.

Comment 9

[Edward Kandrot]

I'll give this an r=kandrot, since it appears to be a simple change (though I
can not verify that it fixes the intial problem).  scc, can I get an sr=?  Thanks.

Comment 10

[Scott Collins]

sr=scc

Comment 11

[Edward Kandrot]

Checked in.  Marking Fixed.