[S390] zcore: Fix reipl device detection
authorMichael Holzheu <holzheu@linux.vnet.ibm.com>
Thu, 22 Apr 2010 15:17:07 +0000 (17:17 +0200)
committerMartin Schwidefsky <sky@mschwide.boeblingen.de.ibm.com>
Thu, 22 Apr 2010 15:17:19 +0000 (17:17 +0200)
The reipl device information is passed from the kernel to zfcpdump
using a pointer in the lowcore (0xe00) that points to the reipl
information Currently if that pointer is not zero, we copy the reipl
information. If the pointer is not initialized and points outside
the accessible memory, it can happen that the memory copy fails.
In that case we currently stop the initialization of zcore which leads
to a failing kernel dump. The correct behavior is to disable the reipl
after dump and continue with zcore intialization.

Signed-off-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
drivers/s390/char/zcore.c

index 18daf16..7217966 100644 (file)
@@ -638,11 +638,7 @@ static int __init zcore_reipl_init(void)
                rc = memcpy_hsa_kernel(ipl_block, ipib_info.ipib, PAGE_SIZE);
        else
                rc = memcpy_real(ipl_block, (void *) ipib_info.ipib, PAGE_SIZE);
-       if (rc) {
-               free_page((unsigned long) ipl_block);
-               return rc;
-       }
-       if (csum_partial(ipl_block, ipl_block->hdr.len, 0) !=
+       if (rc || csum_partial(ipl_block, ipl_block->hdr.len, 0) !=
            ipib_info.checksum) {
                TRACE("Checksum does not match\n");
                free_page((unsigned long) ipl_block);