Hello all,
A month or so ago I wrote a patch against 2.6 that changed the proc_name
of each ESP variant to its actual module name, since our initrd creation
scripts use the contents of /proc/scsi to determine which modules to
include in the ramdisk. Now this is happening with sym53c8xx/sym53c8xx_2
as well. sym53c8xx_2 is detected as the appropriate module during the
install procedure, but appears in /proc/scsi as sym53c8xx, so the wrong
module is inserted into the ramdisk. This patch, against 2.4, changes
the NAME53C8XX #define to "sym53c8xx_2", which proc_name gets set to.
(It turns out that most of the time one gets lucky. sym53c8xx works in
many of the same cases that sym53c8xx_2 did. But one tester reported
that using sym53c8xx instead of sym53c8xx_2 caused his kernel to panic.
More on that perhaps later.)
Marcelo, please apply :)
Signed-off-by: Joshua Kwan <[email protected]>
--- kernel-source-2.4.27-2.4.27/drivers/scsi/sym53c8xx_2/sym_glue.c~ 2004-10-02 17:35:22.000000000 -0700
+++ kernel-source-2.4.27-2.4.27/drivers/scsi/sym53c8xx_2/sym_glue.c 2004-10-02 17:35:38.000000000 -0700
@@ -55,7 +55,7 @@
#include "sym_glue.h"
#define NAME53C "sym53c"
-#define NAME53C8XX "sym53c8xx"
+#define NAME53C8XX "sym53c8xx_2"
/*
* Simple Wrapper to kernel PCI bus interface.
--
Joshua Kwan