August 2021 - Libguestfs - Libguestfs List Archives

[PATCH nbdkit] cache, cow: Fix data corruption in zero and trim on unaligned tail

by Richard W.M. Jones

Finally I tracked down the data corrupter that I've been searching for for most of this week. I think we will need to backport this patch to quite a few branches upstream and downstream. Rich.

2 years, 9 months

2
2
0 / 0

Attach virtual machine disk on gluster volume via libguestfs (Error "No operating system found")

by "Надежда Мозолина, 119 группа"

Hello! I kinly ask you to help me with attach virtual machine disk via libguestfs. There are 3 compute node with Debian 10. Each node is used as hypervisor based on QEMU/KVM. Libvirt0:amd64 5.0.0-4+deb10u1 Libguestfs0:amd 1:1.40.2-2 I create Virtual Machines with disks on GlusterFS volume. ``` <disk type='network' device='disk'> <driver name='qemu' type='qcow2' cache='none' io='threads' discard='unmap'/> <source protocol='gluster' name='TEST/TEST1'> <host name='localhost' port='24007'/> </source> <target dev='vda' bus='virtio'/> <iotune> <read_bytes_sec>157286400</read_bytes_sec> <write_bytes_sec>104857600</write_bytes_sec> <read_iops_sec>40000</read_iops_sec> <write_iops_sec>25000</write_iops_sec> </iotune> <alias name='virtio-disk0'/> <address type='pci' domain='0x0000' bus='0x00' slot='0x09' function='0x0'/> </disk> ``` I would like to work with guest image by libguestfs. Using guestfish (all rights!): ``` ><fs> add TEST/TEST1 protocol:gluster server:localhost:24007 ><fs> run ><fs> list-filesystems /dev/sda1: ext4 /dev/sda5: swap ``` Unfortunatly, the using libguestfs library (C#) results in error "No operating system found". You can find my code in attached files (Source.cpp). How should I use libguetfs library to attach virtual michine disk correctly? -- Надежда Мозолина

2 years, 9 months

2
1
0 / 0

[PATCH nbdkit 0/7] Miscellaneous virt-v2v changes to cache and cow filters

by Richard W.M. Jones

I've spent some time today trying to optimize modular virt-v2v a bit, and these are some suggested changes to nbdkit which should help. Rich.

2 years, 9 months

3
16
0 / 0

[PATCH virt-v2v] v2v: -o rhv-upload: Enable multi-conn

by Richard W.M. Jones

Hi Nir, The experimental patch attached (for discussion) would enable multi-conn. For current virt-v2v this would have no effect since "qemu-img convert" does not make multi-conn NBD connections. However for modular virt-v2v[1] which uses nbdcopy this would allow nbdcopy to make multiple NBD connections to nbdkit running the plugin. (By default 4 connections). If I understand the rhv-upload-plugin code correctly, there is a pool per NBD connection at the moment, so this would create 4 pools and 4*4 backend connections to imageio. This is possibly not desirable, but I guess we can turn the pool into a global object. (Note each NBD connection has the same (destination_url, host, options) tuple). Naturally I've not actually tested any of this. There are some advantages from the nbdcopy side of things, especially because it is able to query extents on the input side in parallel which seems to be advantageous for inputs with slow extent querying (like VDDK). There may be further advantages on the output side because it would allow us to write data to the RHV upload plugin over four sockets, allowing better use of multiple cores. Rich. [1] https://github.com/rwmjones/virt-v2v/commits/2021-virt-v2v-split

2 years, 9 months

3
12
0 / 0

[PATCH virt-v2v 0/2] v2v: Split plugin functions and enable multi-conn

by Richard W.M. Jones

This patch is independent of https://listman.redhat.com/archives/libguestfs/2021-August/msg00009.html but probably needs to be applied on top of it (in any case I've only tested it on top). This doesn't make things any faster because qemu-img convert cannot use multi-conn (hence why the last patch is separate). But it doesn't seem to have any performance impact, so we may as well go with it. The patch isn't quite final because it breaks the test suite. The test suite relies on a single process running the mock ovirtsdk4 replacement API, because it creates a background thread that runs the phony web service (immitating imageio). I'm going to need to think about how to do it now that there are several different scripts involved. However I did test it against real RHV 4.4 and it seems to be OK. Rich.

2 years, 9 months

3
11
0 / 0

[PATCH virt-v2v 0/2] v2v: -o rhv-upload: Generate disk UUIDs

by Richard W.M. Jones

This is a small clean-up/simplification suggested by Nir yesterday. Instead of asking RHV to generate the disk UUIDs in the ! -oo rhv-disk-uuid case, generate them ourselves using "uuidgen -r". Rich.

2 years, 9 months

2
6
0 / 0

[PATCH hivex] lib/node.c: Limit recursion in ri-records (CVE-2021-3622)

by Richard W.M. Jones

Windows Registry hive "ri"-records are arbitrarily nested B-tree-like structures: +-------------+ | ri | |-------------| | nr_offsets | | offset[0] ------> points to another lf/lh/li/ri block | offset[1] ------> | offset[2] ------> +-------------+ It is possible to construct a hive with a very deeply nested tree of ri-records, causing the internal _get_children function to recurse to any depth which can cause programs linked to hivex to crash with a stack overflow. Since it is not thought that deeply nested ri-records occur in real hives, limit recursion depth. If you hit this limit you will see the following error and the operation will return an error instead of crashing: \> ls hivex: _get_children: returning EINVAL because: ri-record nested to depth >= 32 ls: Invalid argument Thanks to Jeremy Galindo for finding and reporting this bug. Reported-by: Jeremy Galindo, Sr Security Engineer, Datto.com Signed-off-by: Richard W.M. Jones <rjones(a)redhat.com> Fixes: CVE-2021-3622 Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1975489 (cherry picked from commit 781a12c4a49dd81365c9c567c5aa5e19e894ba0e) --- lib/node.c | 18 ++++++++++++++---- 1 file changed, 14 insertions(+), 4 deletions(-) diff --git a/lib/node.c b/lib/node.c index 7b002a46..eb7fe93c 100644 --- a/lib/node.c +++ b/lib/node.c @@ -203,7 +203,7 @@ hivex_node_classname (hive_h *h, hive_node_h node) static int _get_children (hive_h *h, hive_node_h blkoff, offset_list *children, offset_list *blocks, - int flags); + int flags, unsigned depth); static int check_child_is_nk_block (hive_h *h, hive_node_h child, int flags); /* Iterate over children (ie. subkeys of a node), returning child @@ -335,7 +335,7 @@ _hivex_get_children (hive_h *h, hive_node_h node, goto error; } - if (_get_children (h, subkey_lf, &children, &blocks, flags) == -1) + if (_get_children (h, subkey_lf, &children, &blocks, flags, 0) == -1) goto error; /* Check the number of children we ended up reading matches @@ -383,7 +383,7 @@ _hivex_get_children (hive_h *h, hive_node_h node, static int _get_children (hive_h *h, hive_node_h blkoff, offset_list *children, offset_list *blocks, - int flags) + int flags, unsigned depth) { /* Add this intermediate block. */ if (_hivex_add_to_offset_list (blocks, blkoff) == -1) @@ -486,7 +486,17 @@ _get_children (hive_h *h, hive_node_h blkoff, } } - if (_get_children (h, offset, children, blocks, flags) == -1) + /* Although in theory hive ri records might be nested to any + * depth, in practice this is unlikely. Recursing here caused + * CVE-2021-3622. Thus limit the depth we will recurse to + * something small. + */ + if (depth >= 32) { + SET_ERRNO (EINVAL, "ri-record nested to depth >= %u", depth); + return -1; + } + + if (_get_children (h, offset, children, blocks, flags, depth+1) == -1) return -1; } } -- 2.32.0

2 years, 9 months

1
0
0 / 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

Libguestfs August 2021