[PATCH nbdkit] tar as a filter.
For review only, this needs some clean up and more tests. My eyes are going cross-eyed looking at the calculate_offset_of_entry function, so time to take a break ... Rich.
This filter can be used to open tar files. It uses the technique first suggested by Eric Blake here: https://www.redhat.com/archives/libguestfs/2020-July/msg00017.html We suggest that nbdkit-tar-plugin is deprecated in nbdkit 1.26, but it might happen later. --- docs/nbdkit-captive.pod | 4 +- filters/offset/nbdkit-offset-filter.pod | 2 +- filters/tar/nbdkit-tar-filter.pod | 108 +++++++ plugins/tar/nbdkit-tar-plugin.pod | 10 + configure.ac | 2 + filters/tar/Makefile.am | 67 ++++ tests/Makefile.am | 20 +- filters/tar/tar.c | 395 ++++++++++++++++++++++++ tests/test-tar-info.sh | 4 +- tests/test-tar.sh | 2 +- TODO | 8 +- 11 files changed, 603 insertions(+), 19 deletions(-) diff --git a/docs/nbdkit-captive.pod b/docs/nbdkit-captive.pod index 390c2191..09628367 100644 --- a/docs/nbdkit-captive.pod +++ b/docs/nbdkit-captive.pod @@ -106,10 +106,10 @@ If the source suffers from temporary network failures L<nbdkit-retry-filter(1)> may help. To overwrite a file inside an uncompressed tar file (the file being -overwritten must be the same size), use L<nbdkit-tar-plugin(1)> like +overwritten must be the same size), use L<nbdkit-tar-filter(1)> like this: - nbdkit -U - tar tar=data.tar file=disk.img \ + nbdkit -U - file data.tar --filter=tar tar-entry=disk.img \ --run 'qemu-img convert -n disk.img $nbd' =head1 EXIT WITH PARENT diff --git a/filters/offset/nbdkit-offset-filter.pod b/filters/offset/nbdkit-offset-filter.pod index 0fd2761e..19df1742 100644 --- a/filters/offset/nbdkit-offset-filter.pod +++ b/filters/offset/nbdkit-offset-filter.pod @@ -84,7 +84,7 @@ L<nbdkit(1)>, L<nbdkit-file-plugin(1)>, L<nbdkit-filter(3)>, L<nbdkit-partition-filter(1)>, -L<nbdkit-tar-plugin(1)>, +L<nbdkit-tar-filter(1)>, L<nbdkit-truncate-filter(1)>. =head1 AUTHORS diff --git a/filters/tar/nbdkit-tar-filter.pod b/filters/tar/nbdkit-tar-filter.pod new file mode 100644 index 00000000..85968668 --- /dev/null +++ b/filters/tar/nbdkit-tar-filter.pod @@ -0,0 +1,108 @@ +=head1 NAME + +nbdkit-tar-filter - read and write files inside tar files without unpacking + +=head1 SYNOPSIS + + nbdkit file FILENAME.tar --filter=tar tar-entry=PATH_INSIDE_TAR + +=head1 EXAMPLES + +=head2 Serve a single file inside a tarball + + nbdkit file file.tar --filter=tar tar-entry=some/disk.img + guestfish --format=raw -a nbd://localhost + +=head2 Opening a disk image inside an OVA file + +The popular "Open Virtual Appliance" (OVA) format is really an +uncompressed tar file containing (usually) VMDK-format files, so you +could access one file in an OVA like this: + + $ tar tf rhel.ova + rhel.ovf + rhel-disk1.vmdk + rhel.mf + $ nbdkit -r file rhel.ova --filter=tar tar-entry=rhel-disk1.vmdk + $ guestfish --ro --format=vmdk -a nbd://localhost + +In this case the tarball is opened readonly (I<-r> option). The +plugin supports write access, but writing to the VMDK file in the +tarball does not change data checksums stored in other files (the +C<rhel.mf> file in this example), and as these will become incorrect +you probably won't be able to open the file with another tool +afterwards. + +=head2 Open a disk image inside a remote tar file + +You can use other plugins apart from L<nbdkit-file-plugin(1)> to +provide the tar file. For example if the tar file is located on a web +server use: + + nbdkit -r curl https://example.com/file.tar \ + --filter=tar tar-entry=disk.img + +=head2 Open an xz-compressed tar file (read-only) + +This filter cannot handle compressed tar files itself, but you can +combine it with L<nbdkit-xz-filter(1)>: + + nbdkit file filename.tar.xz --filter=tar --filter=xz tar-entry=disk.img + +=head1 DESCRIPTION + +C<nbdkit-tar-filter> is a filter which can read and writes files +inside an uncompressed tar file without unpacking the tar file. + +The tar file is provided by the underlying plugin. You must tell the +filter which entry in the tar file you wish to read and write using +the C<tar-entry> parameter. The C<tar-entry> must exactly match file +name in the tar index. Use C<tar tf filename.tar> to list the index +of a tar file. + +This filter will B<not> work directly on compressed tar files. You +have to combine it with another filter as shown in the example above. + +Use the nbdkit I<-r> flag to open the file readonly. This is the +safest option because it guarantees that the tar file will not be +modified. Without I<-r> writes will modify the tar file. + +The disk image cannot be resized. + +=head1 PARAMETERS + +=over 4 + +=item [B<tar-entry=>]PATH_INSIDE_TAR + +The path of the file inside the tarball to serve. This parameter is +required. It must exactly match the name stored in the tarball, so +use S<C<tar tf filename.tar>> + +=back + +=head1 VERSION + +C<nbdkit-tar-filter> first appeared in nbdkit 1.22. It is derived +from C<nbdkit-tar-plugin> which first appeared in nbdkit 1.2. + +=head1 SEE ALSO + +L<nbdkit(1)>, +L<nbdkit-curl-plugin(1)>, +L<nbdkit-file-plugin(1)>, +L<nbdkit-offset-filter(1)>, +L<nbdkit-plugin(3)>, +L<nbdkit-ssh-plugin(1)>, +L<nbdkit-xz-filter(1)>, +L<tar(1)>. + +=head1 AUTHORS + +Richard W.M. Jones. + +Based on the virt-v2v OVA importer written by Tomáš Golembiovský. + +=head1 COPYRIGHT + +Copyright (C) 2017-2020 Red Hat Inc. diff --git a/plugins/tar/nbdkit-tar-plugin.pod b/plugins/tar/nbdkit-tar-plugin.pod index 589e114b..b576d568 100644 --- a/plugins/tar/nbdkit-tar-plugin.pod +++ b/plugins/tar/nbdkit-tar-plugin.pod @@ -6,6 +6,15 @@ nbdkit-tar-plugin - read and write files inside tar files without unpacking nbdkit tar [tar=]FILENAME.tar file=PATH_INSIDE_TAR +=head1 DEPRECATED + +B<The tar plugin is deprecated in S<nbdkit E<ge> 1.22.16> and will be +removed in S<nbdkit 1.26>>. It has been replaced with a filter with +the same functionality, see L<nbdkit-tar-filter(1)>. You can use the +filter like this: + + nbdkit file FILENAME.tar --filter=tar tar-entry=PATH_INSIDE_TAR + =head1 EXAMPLES =head2 Serve a single file inside a tarball @@ -113,6 +122,7 @@ C<nbdkit-tar-plugin> first appeared in nbdkit 1.2. L<nbdkit(1)>, L<nbdkit-offset-filter(1)>, L<nbdkit-plugin(3)>, +L<nbdkit-tar-filter(1)>, L<nbdkit-xz-filter(1)>, L<tar(1)>. diff --git a/configure.ac b/configure.ac index 899a9dcc..b360d863 100644 --- a/configure.ac +++ b/configure.ac @@ -122,6 +122,7 @@ filters="\ retry \ stats \ swab \ + tar \ truncate \ xz \ " @@ -1161,6 +1162,7 @@ AC_CONFIG_FILES([Makefile filters/retry/Makefile filters/stats/Makefile filters/swab/Makefile + filters/tar/Makefile filters/truncate/Makefile filters/xz/Makefile fuzzing/Makefile diff --git a/filters/tar/Makefile.am b/filters/tar/Makefile.am new file mode 100644 index 00000000..43446c88 --- /dev/null +++ b/filters/tar/Makefile.am @@ -0,0 +1,67 @@ +# nbdkit +# Copyright (C) 2019-2020 Red Hat Inc. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions are +# met: +# +# * Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# +# * Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in the +# documentation and/or other materials provided with the distribution. +# +# * Neither the name of Red Hat nor the names of its contributors may be +# used to endorse or promote products derived from this software without +# specific prior written permission. +# +# THIS SOFTWARE IS PROVIDED BY RED HAT AND CONTRIBUTORS ''AS IS'' AND +# ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, +# THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A +# PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL RED HAT OR +# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF +# USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND +# ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT +# OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF +# SUCH DAMAGE. + +include $(top_srcdir)/common-rules.mk + +EXTRA_DIST = nbdkit-tar-filter.pod + +filter_LTLIBRARIES = nbdkit-tar-filter.la + +nbdkit_tar_filter_la_SOURCES = \ + tar.c \ + $(top_srcdir)/include/nbdkit-filter.h \ + $(NULL) + +nbdkit_tar_filter_la_CPPFLAGS = \ + -I$(top_srcdir)/include \ + -I$(top_srcdir)/common/include \ + -I$(top_srcdir)/common/utils \ + $(NULL) +nbdkit_tar_filter_la_CFLAGS = $(WARNINGS_CFLAGS) +nbdkit_tar_filter_la_LDFLAGS = \ + -module -avoid-version -shared $(SHARED_LDFLAGS) \ + -Wl,--version-script=$(top_srcdir)/filters/filters.syms \ + $(NULL) +nbdkit_tar_filter_la_LIBADD = \ + $(top_builddir)/common/utils/libutils.la \ + $(NULL) + +if HAVE_POD + +man_MANS = nbdkit-tar-filter.1 +CLEANFILES += $(man_MANS) + +nbdkit-tar-filter.1: nbdkit-tar-filter.pod + $(PODWRAPPER) --section=1 --man $@ \ + --html $(top_builddir)/html/$@.html \ + $< + +endif HAVE_POD diff --git a/tests/Makefile.am b/tests/Makefile.am index 062ade84..2bc10d70 100644 --- a/tests/Makefile.am +++ b/tests/Makefile.am @@ -830,16 +830,6 @@ test_streaming_SOURCES = test-streaming.c test_streaming_CFLAGS = $(WARNINGS_CFLAGS) $(LIBNBD_CFLAGS) test_streaming_LDADD = $(LIBNBD_LIBS) -# tar plugin test. -TESTS += \ - test-tar.sh \ - test-tar-info.sh \ - $(NULL) -EXTRA_DIST += \ - test-tar.sh \ - test-tar-info.sh \ - $(NULL) - # tmpdisk plugin test. LIBGUESTFS_TESTS += test-tmpdisk TESTS += test-tmpdisk-command.sh @@ -1379,6 +1369,16 @@ EXTRA_DIST += \ test-swab-64w.sh \ $(NULL) +# tar filter test. +TESTS += \ + test-tar.sh \ + test-tar-info.sh \ + $(NULL) +EXTRA_DIST += \ + test-tar.sh \ + test-tar-info.sh \ + $(NULL) + # truncate filter tests. TESTS += \ test-truncate1.sh \ diff --git a/filters/tar/tar.c b/filters/tar/tar.c new file mode 100644 index 00000000..8d6ced04 --- /dev/null +++ b/filters/tar/tar.c @@ -0,0 +1,395 @@ +/* nbdkit + * Copyright (C) 2018-2020 Red Hat Inc. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions are + * met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * * Neither the name of Red Hat nor the names of its contributors may be + * used to endorse or promote products derived from this software without + * specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY RED HAT AND CONTRIBUTORS ''AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, + * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A + * PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL RED HAT OR + * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF + * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND + * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, + * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT + * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +#include <config.h> + +#include <stdio.h> +#include <stdlib.h> +#include <stdbool.h> +#include <stdint.h> +#include <inttypes.h> +#include <string.h> +#include <unistd.h> +#include <poll.h> +#include <assert.h> +#include <sys/types.h> +#include <sys/stat.h> + +#include <pthread.h> + +#include <nbdkit-filter.h> + +#include "cleanup.h" +#include "minmax.h" +#include "utils.h" + +static const char *entry; /* File within tar (tar-entry=...) */ + +/* Offset and size within tarball. + * + * These are calculated once in the first connection that calls + * tar_prepare. They are protected by the lock. + */ +static pthread_mutex_t lock = PTHREAD_MUTEX_INITIALIZER; +static bool offset_initialized = false; +static uint64_t offset, size; + +static int +tar_config (nbdkit_next_config *next, void *nxdata, + const char *key, const char *value) +{ + if (strcmp (key, "tar-entry") == 0) { + if (entry) { + nbdkit_error ("only one tar-entry parameter can be given"); + return -1; + } + entry = value; + return 0; + } + + return next (nxdata, key, value); +} + +static int +tar_config_complete (nbdkit_next_config_complete *next, void *nxdata) +{ + if (entry == NULL) { + nbdkit_error ("you must supply the tar-entry=<FILENAME> parameter"); + return -1; + } + + return next (nxdata); +} + +#define tar_config_help \ + "tar-entry=<FILENAME> (required) The path inside the tar file to serve." + +static int +tar_thread_model (void) +{ + return NBDKIT_THREAD_MODEL_PARALLEL; +} + +struct handle { + /* These are copied from the globals during tar_prepare, so that we + * don't have to keep grabbing the lock on each request. + */ + uint64_t offset, size; +}; + +static void * +tar_open (nbdkit_next_open *next, nbdkit_backend *nxdata, int readonly) +{ + struct handle *h; + + if (next (nxdata, readonly) == -1) + return NULL; + + h = calloc (1, sizeof *h); + if (h == NULL) { + nbdkit_error ("calloc: %m"); + return NULL; + } + return h; +} + +static void +tar_close (void *handle) +{ + free (handle); +} + +/* Calculate the offset of the entry within the tarball. This is + * called with the lock held. The method used is described here: + * https://www.redhat.com/archives/libguestfs/2020-July/msg00017.html + */ +static int +calculate_offset_of_entry (struct nbdkit_next_ops *next_ops, void *nxdata) +{ + const size_t bufsize = 65536; + char output[] = "/tmp/tarXXXXXX"; + int fd; + FILE *fp; + CLEANUP_FREE char *cmd = NULL; + size_t cmdlen = 0; + CLEANUP_FREE char *buf = NULL; + int64_t i, copysize; + bool scanned_ok = false; + + assert (entry); + + /* Temporary file to capture the output from the tar command. */ + fd = mkstemp (output); + if (fd == -1) { + nbdkit_error ("mkstemp: %m"); + return -1; + } + close (fd); + + /* Construct the tar command to examine the tar file. */ + fp = open_memstream (&cmd, &cmdlen); + if (fp == NULL) { + nbdkit_error ("open_memstream: %m"); + return -1; + } + fprintf (fp, "LANG=C tar --no-auto-compress -tRvf - "); + shell_quote (entry, fp); + fprintf (fp, " > "); + shell_quote (output, fp); + if (fclose (fp) == EOF) { + nbdkit_error ("memstream failed: %m"); + return -1; + } + + /* Prepare the copy buffer and copy size. */ + buf = malloc (bufsize); + if (buf == NULL) { + nbdkit_error ("malloc: %m"); + return -1; + } + copysize = next_ops->get_size (nxdata); + if (copysize == -1) + return -1; + + /* Run the tar command. */ + nbdkit_debug ("%s", cmd); + fp = popen (cmd, "w"); + if (fp == NULL) { + nbdkit_error ("tar: %m"); + return -1; + } + + /* Now loop, writing data from the plugin (the tar file) until we + * detect that tar has written something to the output file or we + * run out of plugin. We're making the assumption that the plugin + * is not going to be sparse, which is probably true of most tar + * files. + */ + for (i = 0; i < copysize; i += bufsize) { + int err, r; + const int64_t count = MIN (bufsize, copysize-i); + int64_t j; + struct stat statbuf; + + r = next_ops->pread (nxdata, buf, count, i, 0, &err); + if (r == -1) { + errno = err; + nbdkit_error ("pread: %m"); + pclose (fp); + return -1; + } + for (j = 0; j < count;) { + size_t written = fwrite (&buf[j], 1, count-j, fp); + if (written == 0) { + nbdkit_error ("tar: error writing to subprocess"); + pclose (fp); + return -1; + } + j += written; + } + + /* Did we get something in the output file yet? */ + if (stat (output, &statbuf) == 0 && statbuf.st_size > 0) + break; + } + pclose (fp); + + /* Open the tar output and try to parse it. */ + fp = fopen (output, "r"); + if (fp == NULL) { + nbdkit_error ("%s: %m", output); + return -1; + } + scanned_ok = fscanf (fp, "block %" SCNu64 ": %*s %*s %" SCNu64, + &offset, &size) == 2; + if (fclose (fp) != 0) { + nbdkit_error ("tar subcommand failed, " + "check that the file really exists in the tarball"); + return -1; + } + + unlink (output); + + if (!scanned_ok) { + nbdkit_error ("unexpected output from the tar subcommand"); + return -1; + } + + /* Adjust the offset: Add 1 for the tar header, then multiply by the + * block size. + */ + offset = (offset+1) * 512; + + nbdkit_debug ("tar: offset %" PRIu64 ", size %" PRIu64, offset, size); + + /* Check it looks sensible. XXX We ought to check it doesn't exceed + * the size of the tar file. + */ + if (offset >= INT64_MAX || size >= INT64_MAX) { + nbdkit_error ("internal error: calculated offset and size are wrong"); + return -1; + } + + offset_initialized = true; + + return 0; +} + +static int +tar_prepare (struct nbdkit_next_ops *next_ops, void *nxdata, + void *handle, int readonly) +{ + struct handle *h = handle; + ACQUIRE_LOCK_FOR_CURRENT_SCOPE (&lock); + + if (!offset_initialized) { + if (calculate_offset_of_entry (next_ops, nxdata) == -1) + return -1; + } + + assert (offset_initialized); + assert (offset > 0); + h->offset = offset; + h->size = size; + return 0; +} + +/* Get the file size. */ +static int64_t +tar_get_size (struct nbdkit_next_ops *next_ops, void *nxdata, + void *handle) +{ + struct handle *h = handle; + return h->size; +} + +/* Read data from the file. */ +static int +tar_pread (struct nbdkit_next_ops *next_ops, void *nxdata, + void *handle, void *buf, uint32_t count, uint64_t offs, + uint32_t flags, int *err) +{ + struct handle *h = handle; + return next_ops->pread (nxdata, buf, count, offs + h->offset, flags, err); +} + +/* Write data to the file. */ +static int +tar_pwrite (struct nbdkit_next_ops *next_ops, void *nxdata, + void *handle, const void *buf, uint32_t count, uint64_t offs, + uint32_t flags, int *err) +{ + struct handle *h = handle; + return next_ops->pwrite (nxdata, buf, count, offs + h->offset, flags, err); +} + +/* Trim data. */ +static int +tar_trim (struct nbdkit_next_ops *next_ops, void *nxdata, + void *handle, uint32_t count, uint64_t offs, uint32_t flags, + int *err) +{ + struct handle *h = handle; + return next_ops->trim (nxdata, count, offs + h->offset, flags, err); +} + +/* Zero data. */ +static int +tar_zero (struct nbdkit_next_ops *next_ops, void *nxdata, + void *handle, uint32_t count, uint64_t offs, uint32_t flags, + int *err) +{ + struct handle *h = handle; + return next_ops->zero (nxdata, count, offs + h->offset, flags, err); +} + +/* Extents. */ +static int +tar_extents (struct nbdkit_next_ops *next_ops, void *nxdata, + void *handle, uint32_t count, uint64_t offs, uint32_t flags, + struct nbdkit_extents *extents, int *err) +{ + struct handle *h = handle; + size_t i; + CLEANUP_EXTENTS_FREE struct nbdkit_extents *extents2 = NULL; + struct nbdkit_extent e; + + extents2 = nbdkit_extents_new (offs + h->offset, h->offset + h->size); + if (extents2 == NULL) { + *err = errno; + return -1; + } + if (next_ops->extents (nxdata, count, offs + h->offset, + flags, extents2, err) == -1) + return -1; + + for (i = 0; i < nbdkit_extents_count (extents2); ++i) { + e = nbdkit_get_extent (extents2, i); + e.offset -= h->offset; + if (nbdkit_add_extent (extents, e.offset, e.length, e.type) == -1) { + *err = errno; + return -1; + } + } + return 0; +} + +/* Cache data. */ +static int +tar_cache (struct nbdkit_next_ops *next_ops, void *nxdata, + void *handle, uint32_t count, uint64_t offs, uint32_t flags, + int *err) +{ + struct handle *h = handle; + return next_ops->cache (nxdata, count, offs + h->offset, flags, err); +} + +static struct nbdkit_filter filter = { + .name = "tar", + .longname = "nbdkit tar filter", + .config = tar_config, + .config_complete = tar_config_complete, + .config_help = tar_config_help, + .thread_model = tar_thread_model, + .open = tar_open, + .close = tar_close, + .prepare = tar_prepare, + .get_size = tar_get_size, + .pread = tar_pread, + .pwrite = tar_pwrite, + .trim = tar_trim, + .zero = tar_zero, + .extents = tar_extents, + .cache = tar_cache, +}; + +NBDKIT_REGISTER_FILTER(filter) diff --git a/tests/test-tar-info.sh b/tests/test-tar-info.sh index efaa8ec8..459e5d6f 100755 --- a/tests/test-tar-info.sh +++ b/tests/test-tar-info.sh @@ -56,7 +56,9 @@ qemu-img convert -f raw disk -O qcow2 $disk tar cf $tar $disk # Run nbdkit. -nbdkit -U - tar $tar file=$disk --run 'qemu-img info --output=json $nbd' > $out +nbdkit -U - file $tar \ + --filter=tar tar-entry=$disk \ + --run 'qemu-img info --output=json $nbd' > $out cat $out # Check various fields in the input. diff --git a/tests/test-tar.sh b/tests/test-tar.sh index 3164b826..8f6422cf 100755 --- a/tests/test-tar.sh +++ b/tests/test-tar.sh @@ -49,7 +49,7 @@ tar cf tar.tar test-tar.sh Makefile disk Makefile.am tar tvvf tar.tar # Run nbdkit. -start_nbdkit -P tar.pid -U $sock tar tar=tar.tar file=disk +start_nbdkit -P tar.pid -U $sock file tar.tar --filter=tar tar-entry=disk # Now see if we can open, read and write the disk from the tar file. guestfish -x --format=raw -a "nbd://?socket=$sock" -m /dev/sda1 <<EOF diff --git a/TODO b/TODO index d96b11e7..7332a9eb 100644 --- a/TODO +++ b/TODO @@ -169,12 +169,8 @@ Rust: Suggestions for filters ----------------------- -* tar plugin should really be a filter - * gzip plugin should really be a filter -* libarchive could be used to implement a general tar/zip filter - * LUKS encrypt/decrypt filter, bonus points if compatible with qemu LUKS-encrypted disk images @@ -202,6 +198,10 @@ Suggestions for filters * nbdkit-swab-filter should be able to swap 32 and 64 bits. +* nbdkit-tar-filter should let you specify the tar command, eg. for + platforms like FreeBSD where tar != GNU tar but GNU tar can be + installed under another name. + nbdkit-rate-filter: * allow other kinds of traffic shaping such as VBR -- 2.27.0
On Tue, Jul 07, 2020 at 06:36:42PM +0100, Richard W.M. Jones wrote:
+/* Get the file size. */ +static int64_t +tar_get_size (struct nbdkit_next_ops *next_ops, void *nxdata, + void *handle) +{ + struct handle *h = handle; + return h->size; +}
Interesting crash caused here. I'm going to document this on the public mailing list so we can find it in future, but I do not offer a fix. I noticed during testing that the filter segfaulted if you used something like: nbdkit --filter=tar --filter=xz file disk.tar.xz tar-entry=disk and did a simple query of the export TWICE: qemu-img info nbd:localhost:10809 qemu-img info nbd:localhost:10809 Notice what I'm saying here is that the first qemu-img command worked and the second one caused nbdkit to segfault. The segfault turned out to be this assertion failure: 263 assert (h->exportsize <= INT64_MAX); /* Guaranteed by negotiation phase */ with this stack trace: #1 0x00007fd737780884 in abort () from /lib64/libc.so.6 #2 0x00007fd737780769 in __assert_fail_base.cold () from /lib64/libc.so.6 #3 0x00007fd73778ff86 in __assert_fail () from /lib64/libc.so.6 #4 0x00000000004057b0 in backend_valid_range (b=0x85dbf20, offset=512, count=512) at backend.c:263 #5 0x0000000000406488 in backend_pread (b=0x85dbf20, buf=0x85ec730, count=512, offset=512, flags=0, err=0x7fd736afaa78) at backend.c:478 #6 0x00007fd7372b3b94 in tar_pread (next_ops=0x424520 <next_ops>, nxdata=0x85dbf20, handle=0x85a3130, buf=0x85ec730, count=512, offs=0, flags=0, err=0x7fd736afaa78) at tar.c:315 #7 0x000000000040bdbc in filter_pread (b=0x85da790, handle=0x85a3130, buf=0x85ec730, count=512, offset=0, flags=0, err=0x7fd736afaa78) at filters.c:441 #8 0x0000000000406536 in backend_pread (b=0x85da790, buf=0x85ec730, count=512, offset=0, flags=0, err=0x7fd736afaa78) at backend.c:483 #9 0x00000000004119ad in handle_request (cmd=0, flags=0, offset=0, count=512, buf=0x85ec730, extents=0x0) at protocol.c:241 #10 0x0000000000412bf8 in protocol_recv_request_send_reply () at protocol.c:713 #11 0x0000000000408356 in handle_single_connection (sockin=7, sockout=7) at connections.c:175 #12 0x000000000041820f in start_thread (datav=0x85a1fc0) at sockets.c:337 #13 0x00007fd73792f479 in start_thread () from /lib64/libpthread.so.0 #14 0x00007fd73785cb53 in clone () from /lib64/libc.so.6 The problem is that the xz backend’s exportsize is not cached from a previous call to xz_get_size(), so when we come to try to validate the range the assert fails. The reason for this is tar_prepare only calls the underlying next_ops->get_size() on the very first connection, because on subsequent connections the offset/size has already been cached so we don't need to do the work in calculate_offset_of_entry, and calculate_offset_of_entry is what calls next_ops->get_size. Also tar_get_size doesn't call next_ops->get_size because it doesn't need the result, which would have been the second opportunity to cache exportsize. If neither of these things happen then the core server will crash. I have fixed this in my copy of the tar filter by calling next_ops->get_size in tar_get_size even though the result is not needed. I checked the other plugins and the suspicious ones were: - ext2 - partition - truncate - xz However I was not able to produce a test case that crashed any of them. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com Fedora Windows cross-compiler. Compile Windows programs, test, and build Windows installers. Over 100 libraries supported. http://fedoraproject.org/wiki/MinGW
I have fixed this as far as I can and pushed it, but there are likely still bugs, so please test it! You can now do stuff like: $ nbdkit -f -v curl http://example.com/qcow2.tar.xz \ --filter=tar --filter=xz tar-entry=disk.qcow2 Because the disk inside the tarball in this case is a qcow2 file, it appears as qcow2 on the wire, so: $ guestfish --ro --format=qcow2 -a nbd://localhost Welcome to guestfish, the guest filesystem shell for editing virtual machine filesystems and disk images. Type: ‘help’ for help on commands ‘man’ to read the manual ‘quit’ to quit the shell
<fs> run <fs> list-filesystems /dev/sda1: ext2 <fs> mount /dev/sda1 / <fs> ll / total 19 drwxr-xr-x 3 root root 1024 Jul 6 20:03 . drwxr-xr-x 19 root root 4096 Jul 9 11:01 .. -rw-rw-r--. 1 1000 1000 11 Jul 6 20:03 hello.txt drwx------ 2 root root 12288 Jul 6 20:03 lost+found
The (written in C) nbdkit-tar-plugin still exists. I added a note to the man page that we may remove this in nbdkit 1.26 which is a comfortably long time in the future. But perhaps we won't remove it if it still has advantages such as performance, more features or stability? We'll see. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com virt-df lists disk usage of guests without needing to install any software inside the virtual machine. Supports Linux and Windows. http://people.redhat.com/~rjones/virt-df/
participants (1)
-
Richard W.M. Jones