610283 - ntgc.c: typo in assignment

Status: RESOLVED FIXED

(NSPR :: NSPR, defect)

Description

[Nicolas Kaiser]

User-Agent:       Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.1.15) Gecko/20101106 Gentoo/2.0.10 SeaMonkey/2.0.10
Build Identifier: Mercurial

Looks like a typo to me. On line 57, the assignment is
	t->md.gcContext[3] = context.Edx;
So t->md.gcContext[3] would also match context.Edx in the comment.

Best regards,
Nicolas Kaiser
---
--- a/comm-central/mozilla/nsprpub/pr/src/md/windows/ntgc.c	2010-11-06 13:30:15.913150747 +0100
+++ b/comm-central/mozilla/nsprpub/pr/src/md/windows/ntgc.c	2010-11-08 00:37:19.235639713 +0100
@@ -87,7 +87,7 @@ PRWord *_MD_HomeGCRegisters(PRThread *t,
         t->md.gcContext[0] = 0;                /* context.Eax */
         t->md.gcContext[1] = fiberData[0x2e];  /* context.Ebx */
         t->md.gcContext[2] = 0;                /* context.Ecx */
-        t->md.gcContext[2] = 0;                /* context.Edx */
+        t->md.gcContext[3] = 0;                /* context.Edx */
         t->md.gcContext[4] = fiberData[0x2d];  /* context.Esi */
         t->md.gcContext[5] = fiberData[0x2c];  /* context.Edi */
         t->md.gcContext[6] = fiberData[0x36];  /* context.Esp */

Reproducible: Always

Comment 1

[Wan-Teh Chang]

Nicolas: thanks for the bug report and the patch.  I think you're right.  Note that the file ntgc.c is not being used right now.

Comment 2

[Wan-Teh Chang]

Patch checked in on the NSPR trunk (NSPR 4.8.7).

Checking in ntgc.c;
/cvsroot/mozilla/nsprpub/pr/src/md/windows/ntgc.c,v  <--  ntgc.c
new revision: 3.8; previous revision: 3.7
done

As far as I can tell, we zero the entire PRThread structure
when we allocate it:

http://bonsai.mozilla.org/cvsblame.cgi?file=mozilla/nsprpub/pr/src/threads/combined/pruthr.c&rev=3.41&mark=1005#992

http://bonsai.mozilla.org/cvsblame.cgi?file=mozilla/nsprpub/pr/src/threads/combined/pruthr.c&rev=3.41&mark=1250,1260,1269#1240

http://bonsai.mozilla.org/cvsblame.cgi?file=mozilla/nsprpub/pr/src/md/windows/ntthread.c&rev=3.18&mark=422#413

So this typo should be a cosmetic problem.