Hi Jate,
Jate Sujjavanich wrote:
While debugging the dreaded "bad frame format", I realized that the bus
error handler for m68knommu/Coldfire could be improved.
This patch adds more verbose output for a bus error on Coldfire.
Signed-off-by: Jate Sujjavanich <[EMAIL PROTECTED]>
I have not had a chance to test it, but I believe that the attached
similar patch could also be made to arch/m68knommu/kernel/traps.c for
the 2.6.x series kernel. Do you think the CONFIG_COLDFIRE's are even
needed?
I noticed that in the 2.4 traps.c, there were a lot of ifdef's of code
for 680x0 processors. I have a feeling that these left over from m68k
during the original port. Could these be removed?
I modified the 2.6.x patch a little so that it looks like:
diff -Naurp
/home/gerg/new-wave.2625/ORG.linux-2.6.25-rc1/arch/m68knommu/kernel/traps.c
traps.c
---
/home/gerg/new-wave.2625/ORG.linux-2.6.25-rc1/arch/m68knommu/kernel/traps.c2008-05-06
14:45:05.000000000 +1000
+++ traps.c 2008-05-07 13:14:22.000000000 +1000
@@ -86,6 +86,37 @@ void die_if_kernel(char *str, struct pt_
do_exit(SIGSEGV);
}
+#if defined(CONFIG_COLDFIRE)
+static const char *accesserror_type[] = {
+ /* 0 */ "Not an access or address error nor an interrupted
debug service routine",
+ /* 1 */ "Reserved",
+ /* 2 */ "Interrupt during a debug service routine for faults
other than access errors",
+ /* 3 */ "Reserved",
+ /* 4 */ "Error on instruction fetch",
+ /* 5 */ "TLB miss on opword of instruction fetch",
+ /* 6 */ "TLB miss on extension word of instruction fetch",
+ /* 7 */ "IFP access error while executing in emulator mode",
+ /* 8 */ "Error on data write",
+ /* 9 */ "Error on attempted write to write-protected space",
+ /* 10 */ "TLB miss on data write",
+ /* 11 */ "Reserved",
+ /* 12 */ "Error on data read",
+ /* 13 */ "Attempted read, read-modify-write of protected space",
+ /* 14 */ "TLB miss on data read, or read-modify-write",
+ /* 15 */ "OEP access error while executing in emulator mode"
+};
+
+static void access_errorCF(struct frame *fp)
+{
+ unsigned int vector = fp->ptregs.vector;
+ unsigned int fs;
+
+ fs = ((vector & 0x0c00) >> 8) | (vector & 0x0003);
+ printk(KERN_ERR "Access Error Exception %d: %s\n", fs,
accesserror_type[fs]);
+ force_sig(SIGSEGV, current);
+}
+#endif /* CONFIG_COLDFIRE */
+
asmlinkage void buserr_c(struct frame *fp)
{
/* Only set esp0 if coming from user mode */
@@ -96,6 +127,11 @@ asmlinkage void buserr_c(struct frame *f
printk (KERN_DEBUG "*** Bus Error *** Format is %x\n",
fp->ptregs.format);
#endif
+#if defined(CONFIG_COLDFIRE)
+ if (fp->ptregs.format == 4)
+ access_errorCF(fp);
+#endif
+
die_if_kernel("bad frame format",&fp->ptregs,0);
#if defined(DEBUG)
printk(KERN_DEBUG "Unknown SIGSEGV - 4\n");
This conforms more closely to the kernel formatting guidelines.
I simplified the accesserror_type() to simply report the error
and let buserr_c() handle the die_if_kernel() as it did before too.
Does that still look alright to you?
Regards
Greg
------------------------------------------------------------------------
Greg Ungerer -- Chief Software Dude EMAIL: [EMAIL PROTECTED]
Secure Computing Corporation PHONE: +61 7 3435 2888
825 Stanley St, FAX: +61 7 3891 3630
Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com
_______________________________________________
uClinux-dev mailing list
uClinux-dev@uclinux.org
http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
This message was resent by uclinux-dev@uclinux.org
To unsubscribe see:
http://mailman.uclinux.org/mailman/options/uclinux-dev
