This includes various code cleanups:
(a) A regression test for RHBZ#580246.
(b) Use write instead of fwrite to write out the tar file. This is
just because the error handling of write seems to be better
specified and easier to use.
(c) Use size_t instead of int for length.
(d) Clearer debug messages when in verbose mode.
The problem was this sequence of events:
(1) File transfer goes through OK.
(2) pclose returns failure (because 'tar' subprocess failed)
(3) We try to cancel the transfer by calling cancel_receive.
Step (3) fails because the transfer (as far as the library is
concerned) has succeeded, so causing a hang.
The more fundamental reason why we see steps (1) and (2) is that
'tar' does NOT fail immediately if there is a write error. Instead
it continues reading and discarding the input until the end of the
input before giving "Error exit delayed from previous errors".
IMHO this is a bug with tar, since an ENOSPC write error should
be fatal for tar.
$ guestfish /tmp/disk.img
/tmp/disk.img: unknown command
Did you mean to open a disk image? guestfish -a disk.img
For a list of commands: guestfish -h
For complete documentation: man guestfish
posix_fallocate has a non-standard way to return error indications.
Thus all our calls to posix_fallocate were effectively unchecked. For
example:
$ guestfish alloc test.img 1P
$ echo $?
0
$ ll test.img
-rw-rw-r--. 1 rjones rjones 0 2010-04-06 11:02 test.img
$ rm test.img
With this change, errors are detected and reported properly:
$ ./fish/guestfish alloc test.img 1P
fallocate: File too large
This is a fix for:
https://bugzilla.redhat.com/show_bug.cgi?id=579664
This commit fixes the 'Use%' field in the output so it matches what
coreutils' 'df' command would print.
Firstly we change the calculation to use the space available to root,
not the space available to non-root. This means it matches what 'df'
when run as root in the guest would show.
Secondly we display this rounded up to the next whole percent (ie. using
ceil), which is also what 'df' does.
Thirdly we change the regression test so it tests this.
Note that even with these changes you are not guaranteed to get precisely
the same figures from inside and outside the guest, as it depends on
how quiescent the guest is and how recently the superblock was synced.
Thanks: Rita Wu
Since commit 977edb3185 we have been passing TERM from the
library into the appliance (for the benefit of virt-rescue).
However this changes the output of parted, which I would argue
is a bug in parted:
TERM=xterm /sbin/parted -m -- /dev/sda unit b print 2>&1 | hexdump -C
00000000 1b 5b 3f 31 30 33 34 68 42 59 54 3b 0a 2f 64 65 |.[?1034hBYT;./de|
Notice the escape-sequence junk before the official "BYT;"
header.
By setting TERM=dumb we avoid this.
Previously we used newSVpv (str, len), but if len == 0 then
this means Perl tries to calculate the string length using
strlen(3). This is not desirable when we know the length, in
which case we should use newSVpvn instead.
The problem is that mkfs was making an ext2 filesystem,
which later we were checking with e4fsck. e4fsck corrects
an "error" on the filesystem:
/dev/VG/LV: Adding dirhash hint to filesystem.
e4fsck returns 1 (errors corrected) which we were interpreting
as an error return.
Virt-resize is the main contribution here, a program which can
be used to expand and shrink partitions in disk images.
Virt-list-partitions is used as an ancillary tool for planning
resize operations.
This is similar to 'guestfs_dd', but it copies just a fixed
number of bytes from the source to the destination. It's an
error if the source is too short or if the destination is too
small.
This allows programs to work if they just
#include <guestfs.h>
and no other headers. It's not useful in the general
case, but fixes some configure-time tests, particularly
the one for Ruby on OS X.
As far as I can tell, Darwin has no way to check the peer euid
of a _loopback_ TCP socket. This is required for the "null vmchannel"
implementation to work securely.
Therefore disable this - Darwin will use one of the other supported
vmchannel implementations instead.
