convert several raidframe ioctls to be bitsize idempotent so that
they work the same in 32 and 64 bit worlds, allowing netbsd32 to
configure and query raid properly. remove useless 'row' in a few
places. add COMPAT_80 and put the old ioctls there.
raidframeio.h:
RAIDFRAME_TEST_ACC
- remove, unused
RAIDFRAME_GET_COMPONENT_LABEL
- convert to label not pointer to label
RAIDFRAME_CHECK_RECON_STATUS_EXT
RAIDFRAME_CHECK_PARITYREWRITE_STATUS_EXT
RAIDFRAME_CHECK_COPYBACK_STATUS_EXT
- convert to progress info not pointer to info
RAIDFRAME_GET_INFO
- version entirely.
raidframevar.h:
- rf_recon_req{} has row, flags and raidPtr removed (they're
not a useful part of this interface.)
- RF_Config_s{} and RF_DeviceConfig_s{} have numRow/rows removed.
- RF_RaidDisk_s{} is re-ordered slightly to fix alignment
padding - the actual data was already OK.
- InstallSpareTable() loses row argument
rf_compat32.c has code for RF_Config_s{} in 32 bit mode, used
by RAIDFRAME_CONFIGURE and RAIDFRAME_GET_INFO32.
rf_compat80.c has code for rf_recon_req{}, RF_RaidDisk_s{} and
RF_DeviceConfig_s{} to handle RAIDFRAME_FAIL_DISK,
RAIDFRAME_GET_COMPONENT_LABEL, RAIDFRAME_CHECK_RECON_STATUS_EXT,
RAIDFRAME_CHECK_PARITYREWRITE_STATUS_EXT,
RAIDFRAME_CHECK_COPYBACK_STATUS_EXT, RAIDFRAME_GET_INFO.
move several of the per-ioctl code blocks into separate functions.
add rf_recon_req_internal{} to replace old usage of global
rf_recon_req{} that had unused void * in the structure, ruining
it's 32/64 bit ABI.
add missing case for RAIDFRAME_GET_INFO50.
adjust raid tests to use the new .conf format, and add a case to
test the old method as well.
raidctl:
deal with lack of 'row' members in a couple of places.
fail request no longer takes row.
handle "START array" sections with just "numCol numSpare", ie
no "numRow" specified. for now, generate old-style configuration
but update raidctl.8 to specify the new style (keeping reference
to the old style.)
note that: RF_ComponentLabel_s::{row,num_rows} and
RF_SingleComponent_s::row are obsolete but not removed yet.
1. Don't force use of "for" when "while" works better.
2. No need to check c != '\0' when we also check (c == ' ' || c == '\t')
3. Use the size of the buffer we're using, rather than a different one
(not really a concern, they're the same size)
4. Don't use fscanf() to read file data, use fgets() & sscanf().
5. After using a pointer as a char *, validate alignment before switching
to int * (can only fail if kernel #define gets set stupidly) Or #6...
6. Validate sparemap file name isn't too long for assigned space.
7. recognise that strlen() returns size_t - don't shove it into an int.
8. On out of mem, be more clear which allocation failed in warning msg.
ATF tests all pass. But I don't think they use sparemap files.
Coincidentally, the change also works around a gcc 5.1 bug which causes
a segmentation fault when trying to compile the longer version (guess
the compiler got exhausted, or something).
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66345
Replace atoi(3) by strtol(3), and check that numbers are valid,
positive, and in int32_t range. The previous lack of check could
silently lead to the same serial being set to all RAID volumes
for instance because given numbers were bigger than INT_MAX. The
consequence is in an awful mess when RAIDframe would mix volumes...
component in a failed RAID set in order to avoid potentially configuring
RAId 1 sets with erroneous values taken from random extent data in the
replacement component partitions.
The server must of course have some disks configured. Let's say
we have this simple server with disks as a few sparse host files:
main()
{
rump_init();
rump_pub_etfs_register("/disk1", "./disk1.img", RUMP_ETFS_BLK);
rump_pub_etfs_register("/disk2", "./disk2.img", RUMP_ETFS_BLK);
rump_pub_etfs_register("/disk3", "./disk3.img", RUMP_ETFS_BLK);
rump_pub_etfs_register("/disk4", "./disk4.img", RUMP_ETFS_BLK);
pause();
}
And we run the server:
mainbus0 (root)
Kernelized RAIDframe activated
/disk1: hostpath ./disk1.img (97 GB)
/disk2: hostpath ./disk2.img (97 GB)
/disk3: hostpath ./disk3.img (97 GB)
/disk4: hostpath ./disk4.img (97 GB)
We can then configure the raid against the server:
> ./raidctl -c theraid.conf raid0
And lo, we have evidence of a level1 raid in the server dmesg:
raid0: RAID Level 1
raid0: Components: /disk1 /disk2 /disk3 /disk4
raid0: Total Sectors: 409599744 (199999 MB)
yea, i initialized it already in a previous run:
> ./raidctl -S raid0
Reconstruction is 100% complete.
Parity Re-write is 100% complete.
Copyback is 100% complete.
"marked clean" after however much inactivity; it is *actually* clean
as soon as the component disks all do their thing (on the order of ms,
usually), just the same as before.
The bikeshed is now less of a taupe and more of an ecru.
Drastically reduces the amount of time spent rewriting parity after an
unclean shutdown by keeping better track of which regions might have had
outstanding writes. Enabled by default; can be disabled on a per-set
basis, or tuned, with the new raidctl(8) commands.
Discussed on tech-kern@ to a general air of approval; exhortations to
commit from mrg@, christos@, and others.
Thanks to Google for their sponsorship, oster@ for mentoring the
project, assorted developers for trying very hard to break it, and
probably more I'm forgetting.
component label. raidctl(8) should now print the correct number of
blocks for RAID sets larger than 1TB.
Patch supplied by Bernhard Moellemann in PR bin/40479.
nakayama