+++ This bug was initially created as a clone of Bug #253000 +++ Description of problem: The dmraid format handlers may need to byteswap on-disk data structures while reading or writing. The nvidia handler accomplishes this by defining a to_cpu function when built on these archs. There is a typo in the current version of the code that causes a segfault when probing devices: static void to_cpu(void *meta) { struct nv *nv = meta; unsigned int i = NV_SIGNATURES; struct nv_array_base *array = &nv->array; CVT32(nv->size); CVT32(nv->chksum); CVT16(nv->version); CVT32(nv->capacity); CVT32(nv->sectorSize); CVT32(nv->unitFlags); CVT32(array->version); while (i--); <-- CVT32(array->signature[i]); That semicolon causes us to exit the loop with i == 2^32 -1 and segfault on the access to signatures[i], e.g: (gdb) run -i -a y Starting program: /usr/src/redhat/BUILD/dmraid/1.0.0.rc14/tools/dmraid -i -a y Program received signal SIGSEGV, Segmentation fault. 0x0000000080016eba in to_cpu (meta=Variable "meta" is not available. ) at ./datastruct/byteorder.h:32 32 { (gdb) bt #0 0x0000000080016eba in to_cpu (meta=Variable "meta" is not available. ) at ./datastruct/byteorder.h:32 #1 0x0000000080009d9a in read_raid_dev (lc=0x80030270, di=0x800305c0, f_read_metadata=Variable "f_read_metadata" is not available. ) at format/format.c:433 #2 0x000000008001746e in nv_read (lc=Variable "lc" is not available. ) at format/ataraid/nv.c:208 #3 0x000000008000ba7e in dmraid_read (lc=0x80030270, di=0x800305c0, format=0x0, type=FMT_RAID) at metadata/metadata.c:596 #4 0x000000008000bdcc in discover_raid_devices (lc=0x80030270, devices=0x0) at metadata/metadata.c:700 #5 0x0000000080005078 in perform (lc=0x80030270, argv=0x3fffffff9c8) at commands.c:624 #6 0x000000008000438c in main (argc=4, argv=0x3fffffff9c8) at dmraid.c:34 (gdb) quit Attached patch has been tested successfully on s390x & submitted upstream. Version-Release number of selected component (if applicable): dmraid-1.0.0.rc14 How reproducible: 100% Steps to Reproduce: 1. dmraid -ay Actual results: # dmraid -ay Segmentation Fault Expected results: # dmraid -ay No RAID disks Additional info: Low impact on s390 since these don't typically have fakeraid hardware ;) On PPC the problem is probably more important since these can use these types of RAID card.
Created attachment 161665 [details] patch against upstream dmraid-1.0.0.rc14
Problem & fix now also confirmed on ppc/ppc64
Bryn, thanks for the patch.
This request was evaluated by Red Hat Product Management for inclusion in a Red Hat Enterprise Linux maintenance release. Product Management has requested further review of this request by Red Hat Engineering, for potential inclusion in a Red Hat Enterprise Linux Update release for currently deployed products. This request is not yet committed for inclusion in an Update release.
This missed the 5.2 deadline. We will get it checked in early to 5.3, so it will be available as hot fix, as needed.
In 5.3.
An advisory has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on therefore solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHEA-2009-0078.html