host, and vice versa), to fix PR#44203.
Add support for growing (but not yet shrinking) UFS2 file systems. Partially
addresses PR#44205.
While I'm here, reformat the code for closer adherence to KNF.
Fairly extensive testing was performed, using the shortly-to-be-committed
updated ATF tests. Patch posted to tech-userlevel on 21 December 2010,
no comments.
unsupported, while catching up to some changes in my local tree which
will hopefully support them at some time in the future.
Also, change "device" variable to "special", to reflect the fact
that resize_ffs will work on a plain file.
that while modstat and modunload and builtin modules work exactly
the same as in the host case, modload loads file system kernel
modules from the rump kernel namespace. By default, archs which
have rump support for the kernel kernel ABI have the host module
directory mapped into the rump kernel namespace at the same location
(/stand/...). Therefore, if the *host* module directory is populated,
"rump.modload foo" will work as expected. Otherwise, RUMP_MODULEBASE
can be used to point to the module directory.
and depending on file system data, can actually be a false error.
Fixes what I was actually testing for in bin/44209, though the
actual problem was not what I originally described.
copy appropriate data to where they are expected in the updated superblock.
When writing the updated superblock, move the updated values back to the
old ffsv1 superblock locations. Also check for old superblock format when
updating the last cylinder group and adjust cg_old_ncyl appropriately.
Derived from how mksf sets them. Should address PR bin/44209.
to ensure it's been properly extended. Clears up some problems at certain
blocksizes which showed up during creation of atf tests, which is done
using file-backed file systems.
when testing that the last sector of the new size is writeable, make
sure we're ACTUALLY writing in the new space, instead of possibly
overwriting something in the existing fs.
Discovered while writing tests - tests which uncovered file corruption at
certain block sizes.
XXX should rewrite writeat() to expect fs blocks instead of disk blocks.
OK mlelstv@
for KEYGEN_RANDOMKEY.
Print a warning if such a refusal is made---this will help the user understand
why there is an error.
Patch provided by: Taylor R Campbell <campbell+netbsd@mumble.net>.
- sync usage comment with current reality
- sort includes
- wrap lines
- use EXIT_FAILURE consistently
- make error messages consistent: Cannot->Can't
- Remove "Old FFSv1 macros" in favor of system macros in ufs/ffs/fs.h .
Leave dblksize() because it uses the on-disk dinode structure.
More cleanup is needed.
No functional changes intended.
were commented out with XXX and a notation to "fix once fsck is fixed."
fsck seems to have been fixed for this particular issue sometime in the
7 years since the code was brought into the tree.
Update cg_old_niblk instead of cg_ni_blk, since this tool
currently supports ffsv1 only.
With these two changes, I can grow a file system and have the result
be clean according to fsck_ffs. Shrinking still results in an unclean
file system.
OK mhitch@
While I'm here, fix a typo in an error message.
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.
current testing purpose is to create a file system with
block size > MAXPHYS.
(the check doesn't make that much sense anyway in these days of
mobile file systems, since we're interested in MAXPHYS where we
attempt to mount the file system, not where we happen to create it)
This is useful for automated environments where everything (rpcbind,
mountd, nfsd and the client) is started in parallel in a split
second and there is a small chance we will race in there before
everything has been communicated to rpcbind.
we're ELF now, and there are many missing checks against OBJECT_FMT.
if we ever consider switching, the we can figure out what new ones
we need but for now it's just clutter.
this doesn't remove any of the support for exec_aout or any actually
required-for-boot a.out support, only the ability to build a netbsd
release in a.out format. ie, most of this code has been dead for
over a decade.
i've tested builds on vax, amd64, i386, mac68k, macppc, sparc, atari,
amiga, shark, cats, dreamcast, landisk, mmeye and x68k. this covers
the 5 MACHINE_ARCH's affected, and all the other arch code touched.
it also includes some actual run-time testing of sparc, i386 and
shark, and i performed binary comparison upon amiga and x68k as well.
some minor details relevant:
- move shlib.[ch] from ld.aout_so into ldconfig proper, and cut them
down to only the parts ldconfig needs
- remove various unused source files
- switch amiga bootblocks to using elf2bb.h instead of aout2bb.h
tells whether it should detect and convert to binary a hexadecimal octet
string of the form 0x0123ABab, or leave those strings undecoded.
If the argument for a 'media', 'mediamode', 'mediaopt', '-mediaopt',
'nwkey', or 'bssid' keyword is a hexadecimal octet string, do not detect
and decode it. (Note that setifnwkey decodes hexadecimal strings on its
own.)
This fixes a bug noticed by Jim Miller where the trailing zero-octets
were discarded from hexadecimal octet-string arguments for 'nwkey'.
a live file system.
While here modify snap_open() to accept a character device as its
first arg and remove now unneeded get_snap_device().
Reviewed by: Manuel Bouyer <bouyer@netbsd.org>
destroy / create to fail. When reading the GPT label from the end of the disk
ignore errors if the GPT label at the beginning of the disk was not found.
ramdisks and prefer disklabel elsewhere.
Based on discussion on affected port lists (port-sparc port-sparc64
port-sun3 port-sun2 port-atari port-mvme68k).
All listed ports plus amd64 test built after change
- drop the notion of frags (LFS fragments) vs fsb (FFS fragments)
The code uses a complicated unity function that just makes the
code difficult to understand.
- support larger sector sizes. Fix disk address computations
to use DEV_BSIZE in the kernel as required by device drivers
and to use sector sizes in userland.
- Fix several locking bugs in lfs_bio.c and lfs_subr.c.
In particular:
- newfs will not try to erase the label
- fsck_ffs will not try to validate the label
This lets newfs and fsck work on 2048-byte-per-sector media.
Does Apple UFS support such media and how?
uint64_t usermem. This only becomes relevant if you have several TB of RAM.
Promoting cachebufs to uint64_t is not necessary as it gets limited to
(currently) 512 anyway.
fixes the last issue of PR: 19852
information to atactl identify output.
Also:
- remove caddr_t cast
- warn about invalid IDENTIFY data checksum (when possible)
- humanize capacity in power-of-10 format
- remove semi-pointless ATAPI check
- slightly rework command queue depth output to be less conversational
(treating a target disk as a regular file and suppressing ioctl(2)s)
on reading/writing disklabel in a target file.
This allows cross build enviroment creating bootable disk images
for targets in different endian.
No functional changes to native (non-tools) disklabel(8) command.
Closes PR toolchain/42357.
Info can be specified with -A parameter.
Default is based on how the first partition is defined.
For empty disks larger than 128GB (arbitrary figure) use 1MB alignment.
Rename (with #defines) the variables use for aligning partitions to
separate them from the bios geometry.
All in advance of allowing other partition alignments (eg 2048 sectors).
"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.
>> Allow MB, GB and CYL (not just M, G and C) and lower case.
>> Don't output a splurious 'd' before "cyl".
>> Fixes PR/37414.
XXX "NNcy" is also allowed?
Instead, use proper macro defined in Makefile per ${MACHINE_ARCH}.
__${MACHINE_ARCH}__ doesn't represent an architecture of tool's target
but an architecture of binaries being compiled, so required features
are not prolery enabled or unintentionally enabled on certain host
and target combinations during src/tools build.
- Use %.40g rather than %g when printing sectors and MB for existing
partition size/offset.
Changes [1.93802e+06c, 1953525105s, 953870M]:
to: [1938021c, 1953525105s, 953869.6875M]:
It just fakes MBR partition map which contains 1MB FAT16B partition
and ~1GB OpenBSD partition, and we can always create necessary
MBR partitions for OpenFirmware by the fdisk(8) command itself.
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.
- Separate the suser part of the bsd44 secmodel into its own secmodel
and directory, pending even more cleanups. For revision history
purposes, the original location of the files was
src/sys/secmodel/bsd44/secmodel_bsd44_suser.c
src/sys/secmodel/bsd44/suser.h
- Add a man-page for secmodel_suser(9) and update the one for
secmodel_bsd44(9).
- Add a "secmodel" module class and use it. Userland program and
documentation updated.
- Manage secmodel count (nsecmodels) through the module framework.
This eliminates the need for secmodel_{,de}register() calls in
secmodel code.
- Prepare for secmodel modularization by adding relevant module bits.
The secmodels don't allow auto unload. The bsd44 secmodel depends
on the suser and securelevel secmodels. The overlay secmodel depends
on the bsd44 secmodel. As the module class is only cosmetic, and to
prevent ambiguity, the bsd44 and overlay secmodels are prefixed with
"secmodel_".
- Adapt the overlay secmodel to recent changes (mainly vnode scope).
- Stop using link-sets for the sysctl node(s) creation.
- Keep sysctl variables under nodes of their relevant secmodels. In
other words, don't create duplicates for the suser/securelevel
secmodels under the bsd44 secmodel, as the latter is merely used
for "grouping".
- For the suser and securelevel secmodels, "advertise presence" in
relevant sysctl nodes (sysctl.security.models.{suser,securelevel}).
- Get rid of the LKM preprocessor stuff.
- As secmodels are now modules, there's no need for an explicit call
to secmodel_start(); it's handled by the module framework. That
said, the module framework was adjusted to properly load secmodels
early during system startup.
- Adapt rump to changes: Instead of using empty stubs for securelevel,
simply use the suser secmodel. Also replace secmodel_start() with a
call to secmodel_suser_start().
- 5.99.20.
Testing was done on i386 ("release" build). Spearated module_init()
changes were tested on sparc and sparc64 as well by martin@ (thanks!).
Mailing list reference:
http://mail-index.netbsd.org/tech-kern/2009/09/25/msg006135.html
0 for success; 1 for error; and 2 when no devices of the appropriate type
are defined in fstab(5). Previously, "no apropriate devices" was
indistinguishable from "error".
so that this part doesn't get included in install media versions
of ifconfig, as per comments on source-changes@ from Izumi Tsutsui.
Also re-instate the use of ${.CURDIR} when setting up the include path.
may be included by different parts of the source tree, in particular
deep down in the distrib/ tree, where ../.. isn't sufficient to "climb"
up to the top of the tree. Fixes the build at least for our arc port.
Pfsync interface exposes change in the pf(4) over a pseudo-interface, and can
be used to synchronise different pf.
This work was part of my 2009 GSoC
No objection on tech-net@
kernel refuse to mount a filesystem read-write (booting a system
multiuser with critical filesystems read-only is bad):
Add a check_wapbl() which will check some WAPBL values in the superblock,
and try to read the journal via wapbl_replay_start() if there is one.
pfatal() if one of these fail (abort boot if in preen mode,
as "CONTINUE" otherwise). In non-preen mode the bogus journal will
be cleared.
check_wapbl() is always called if the superblock supports WAPBL.
Even if FS_DOWAPBL is not there, there could be flags asking the
kernel to clear or create a log with bogus values which would cause the
kernel refuse to mount the filesystem.
Discussed in
http://mail-index.netbsd.org/tech-kern/2009/08/17/msg005896.html
and followups.
addresses. Make the kernel support SIOC[SG]IFADDRPREF for IPv6
interface addresses.
In in6ifa_ifpforlinklocal(), consult preference numbers before
making an otherwise arbitrary choice of in6_ifaddr. Otherwise,
preference numbers are *not* consulted by the kernel, but that will
be rather easy for somebody with a little bit of free time to fix.
Please note that setting the preference number for a link-local
IPv6 address does not work right, yet, but that ought to be fixed
soon.
In support of the changes above,
1 Add a method to struct domain for "externalizing" a sockaddr, and
provide an implementation for IPv6. Expect more work in this area: it
may be more proper to say that the IPv6 implementation "internalizes"
a sockaddr. Add sockaddr_externalize().
2 Add a subroutine, sofamily(), that returns a struct socket's address
family or AF_UNSPEC.
3 Make a lot of IPv4-specific code generic, and move it from
sys/netinet/ to sys/net/ for re-use by IPv6 parts of the kernel and
ifconfig(8).
or route(8): it tells ifconfig(8) to try to resolve numbers to
hosts and service names.
This default ifconfig behavior stays the same as it always was.
a struct called kernelops, which contains standard system calls
for the normal case and rump system calls for the rump case.
Make it possible to run the lfs cleaner in a library fashion (taking
the quick route with the implementation).
