Date: Wed, 9 Jun 1999 17:20:51 +0200
From: Marcus Brinkmann <[EMAIL PROTECTED]>
On Wed, Jun 09, 1999 at 05:20:48PM +0200, Per Lundberg wrote:
>
> > maybe you want to apply the same changes to the standalone db packages?
> > You're certainly welcome.
>
> What do mean is the problem in the db packages? AFAIK the dbopen() is a
> glibc call, and should be fixed in glibc.. but I could be wrong.
Ah, no, it may well be that you are correct and me is not. I didn't
look that closely as soon as Mark said he found the problem.
Roland checked in the following patch to libc:
1999-06-07 Roland McGrath <[EMAIL PROTECTED]>
* db2/os/os_oflags.c (__db_oflags): Fix checking of O_ACCMODE bits to
be POSIX compliant. Prior definition was broken for Hurd.
Reported by Mark Kettenis <[EMAIL PROTECTED]>.
Index: os_oflags.c
===================================================================
RCS file: /var/cvsroot/libc/db2/os/os_oflags.c,v
retrieving revision 1.1.1.1
diff -u -r1.1.1.1 os_oflags.c
--- os_oflags.c 1999/05/22 14:51:08 1.1.1.1
+++ os_oflags.c 1999/06/09 21:01:52
@@ -33,16 +33,21 @@
u_int32_t dbflags;
/*
- * XXX
- * Convert POSIX 1003.1 open(2) flags to DB flags. Not an exact
- * science as most POSIX implementations don't have a flag value
- * for O_RDONLY, it's simply the lack of a write flag.
+ * Convert POSIX 1003.1 open(2) flags to DB flags.
*/
dbflags = 0;
+ switch (oflags & O_ACCMODE) {
+ case O_RDONLY:
+ dbflags |= DB_RDONLY;
+ break;
+ case O_WRONLY:
+ case O_RDWR:
+ break;
+ default: /* Bogus flags value from user. */
+ /* XXX no way to return error from here */
+ }
if (oflags & O_CREAT)
dbflags |= DB_CREATE;
- if (!(oflags & (O_RDWR | O_WRONLY)) || oflags & O_RDONLY)
- dbflags |= DB_RDONLY;
if (oflags & O_TRUNC)
dbflags |= DB_TRUNCATE;
return (dbflags);
If debian has a seperate db2 package that is not based on the db2
sources in glibc, but on the upstream sources instead, this patch
should be applied there too.
Mark
