Files in subdirectories of directories that have the nodump flag set
are sometimes incorrectly being dumped.
The problem arises because the subdirectory only gets its entry
cleared from usedinomap if it is also present in dumpinomap, and it is
the absence of a directory in usedinomap that internally indicates
that the directory is under the effects of UF_NODUMP (either directly
or inherited).
FreeBSD PR: 32414
Submitted by: David C Lawrence <tale@dd.org>
to define the tape size. Requires the tape driver to either return ENOSPC
at end of media, or 0 when a write is attempted (such as the "early warning"
support in st(4) enabled with "mt eew 1"). From FreeBSD.
- Fix error if first tape was write protected. Fix NetBSD PRs 4754 and 6098.
- Make dump exit codes confirm to manual page.
- Use \a instead of \7 to make noise.
- Fix estimated number of tapes for huge dumps to cartridges.
- Use <sys/queue.h> SLIST_* instead of home-rolled lists.
- Do not exit if unable to read or create /etc/dumpdates.
- Support output (tape) device returning ENOSPC for end-of-media on a write.
Fixes by me:
- Remove unused ddates_in.
- Don't dump core if SIGINFO is received before 1 second has elapsed.
- Only process SIGINFO in current "active" child.
- Don't dump core in -w if dumpdates wasn't readable and ddatev == NULL
- Minor KNF; wrap some lines
past EOM and fail with "short read error". This check was part of
traverse.c::bread(), and was removed when that function was migrated
to rcache.c::rawread() as part of the "read cache" functionality.
This should fix the problem with dump barfing on "short read" when "-r 0",
but I'm still debugging the problem where dump gets to a point and
"does nothing" when the read cache is being used.
tape when a volume is full, and try to reopen the tape drive for 2 mn.
To be used with tape changers which load the next tape when the current one is
ejected.
While I'm there fix eject handling for remote tape.
va_start() in between; reorder code so that we call vsnprintf and then
just fputs that buffer. crank the size of lastmsg whilst we're here
problem noted by Hideo Saito in [bin/14348].
determine the endianness of the `struct fs *o' superblock from o->fs_magic
and set needswap as necessary, rather than trusting the caller to get
it right. invariably, almost every caller of ffs_sb_swap() was calling it
with ns set to the wrong value for ns anyway!
ansi KNF ffs_bswap.c declarations whilst here.
this fixes all sorts of problems when trying to use other-endian file systems,
notably the kernel trying to access memory *way* off, possibly corrupting or
panicing, and userland programs SEGVing and/or corrupting things (e.g,
"fsck_ffs -B" to swap a file system endianness).
whilst the previous rev of ffs_bswap.c (1.10, 2000/12/23) made this problem
worse, i suspect that the problem was always there and previous versions
just happened not to trash things at the wrong time.
FFS_EI should now be a lot more stable.
- if it's a path to an unmounted file-system listed in /etc/fstab, use
that instead of assuming the user wanted a subtree dump of the parent
directory. this restores the behaviour of dump before the subtree
dumping code went in.
- if it's a path to a mounted file-system which is not in /etc/fstab,
use the info from getmntinfo(3). previously, dump would choke.
* implement error checked malloc(), calloc(), strdup(), and use
appropriately (some of the calloc()s weren't being checked)
* use 'file-system' instead of 'filesystem' in the man page
header. mostly based on [bin/6715] from Brian Grayson <bgrayson@ece.utexas.edu>
old filesystems don't have fs_qbmask and fs_qfmask set in the superblock;
fudge them as per code in /sys/ufs/ffs/ffs_vfsops.c::ffs_oldfscompat().
bug and suggested fix from Robert Elz <kre@munnari.OZ.AU> in [bin/6610]
XXX: there may be other variables required as well...
commit message from Bill Fenner:
Turn on TCP_NODELAY on the remote socket, to turn off sender silly window
syndrome avoidance. The combination of SWS avoidance and ack-every-other
causes low throughput if the block size divided by the MSS is odd (which
is true with the default block size and MSS).
Turning on TCP_NODELAY disables the Nagle algorithm and sender SWS avoidance.
The rdump request/response protocol can not invoke Nagle and cannot cause
SWS, so this has no negative effects.
Also, put back the code that sets the TOS to "throughput", which seems to
have been erroneously removed during the Lite-2 merge.
- ensure hostname from gethostname() is nul-terminated in all cases
- minor KNF
- use MAXHOSTNAMELEN over various other values/defines
- be safe will buffers that hold hostnames
* exit after providing an estimate if -S was given. the PR used -e,
but checking around indicated prior art in Solaris usin -S.
* remove superfluous 'DUMP:' prefix in two messages
* initialise blocksperfile explicitly (not necessary, but everything
else in that section gets initialised, so be consistant :)
* display the ``pretty'' name of the dumped directory, so the user
knows if it's a subset or not
enami