Forever since its introduction the main loop in the function dealing
with multi-vector MSI had error exit points ("break") with different
properties: In one case no IRQ descriptor lock is being held.
Nevertheless the subsequent error cleanup path assumed such a lock would
uniformly need releasing. Identify the case by setting "desc" to NULL,
thus allowing the unlock to be skipped as necessary.
This is CVE-2024-31143 / XSA-458.
Coverity ID: 1605298
Fixes: d1b6d0a024 ("x86: enable multi-vector MSI")
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Roger Pau Monné <roger.pau@citrix.com>
The containers are exactly as per 20.04 (Focal). However, this now brings us
to 5 releases * 4 build jobs worth of Ubuntu testing, which is overkill.
The oldest and newest toolchains are the most likely to find problems with new
code, so reduce the middle 3 releases (18/20/22) to just a single smoke test
each.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
As with 16.04 (Xenial), with python3-setuptools included. Having this package
only in some containers was intentional; see commit bbc72a7877 ("automation:
Add python3's setuptools to some containers") for the rational.
This saves ~500M:
registry.gitlab.com/xen-project/xen/ubuntu 20.04-x86_64 1.06GB
registry.gitlab.com/xen-project/xen/ubuntu focal 1.57GB
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
As with 16.04 (Xenial), except that ninja-build package is available so QEMU
can be built.
This halves the size of the container:
registry.gitlab.com/xen-project/xen/ubuntu 18.04-x86_64 857MB
registry.gitlab.com/xen-project/xen/ubuntu bionic 1.44GB
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Rework the container to be non-root, use heredocs for legibility, and use
apt-get --no-install-recommends to keep the size down.
Ubuntu Xenial has no ninja-build package, so can't build the QEMU referenced
by Xen 4.16. Therefore, drop the dependencies too.
This saves ~500M:
registry.gitlab.com/xen-project/xen/ubuntu 16.04-x86_64 698MB
registry.gitlab.com/xen-project/xen/ubuntu xenial 1.21GB
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
See prior patch for most discussion.
Despite appearing to be a fixed release (and therefore not marked as permitted
failure), the dockerfile references the `leap` tag which is rolling in
practice. Switch to 15.6 explicitly, for better test stability.
Vs tumbleweed, use `zypper update` rather than dist-upgrade, and retain the
RomBIOS dependencies; bin86 and dev86.
In terms of size, this saves ~700M:
registry.gitlab.com/xen-project/xen/opensuse leap-15.6-x86_64 1.33GB
registry.gitlab.com/xen-project/xen/suse opensuse-leap 2.05GB
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Existing as suse:opensuse-tumbleweed is a historical quirk, and adjusted for
consistency with all the other containers.
Make it non-root, use heredocs for legibility, and use the zypper long names
for the benefit of those wondering what was being referenced or duplicated.
Trim the dependencies substantially. Testing docs isn't very interesting and
saves a lot of space. Other savings come from removing a huge pile of
optional QEMU dependencies (QEMU just needs to build the Xen parts to be
useful here, not have a full GUI environment).
Finally, there where some packages such as bc, libssh2-devel, libtasn1-devel
and nasm that I'm not aware of any reason to have had, even historically.
This shaves 1G off the size of the container:
registry.gitlab.com/xen-project/xen/opensuse tumbleweed-x86_64 1.39GB
registry.gitlab.com/xen-project/xen/suse opensuse-tumbleweed 2.35GB
Furthermore, identify which components of the build use which dependencies,
which will help managing them in the future.
Thanks to Olaf Hering for dependency fixes that have been subsumed into this
total overhaul.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
The containers are both much smaller, with stable toolchains over time, and
this at least means we're not doing all testing with a single compiler.
Rename the jobs to follow to sort coherently ($DISTRO-$VERSION-$ARCH-*) and
reposition the jobs to optimise starting the smoke test.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
Reviewed-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
For starters, they're slightly smaller:
$ docker image list <snip>
registry.gitlab.com/xen-project/xen/debian 12-riscv64 772MB
registry.gitlab.com/xen-project/xen/debian 11-riscv64 422MB
registry.gitlab.com/xen-project/xen/archlinux current-riscv64 2.32GB
They also not rolling distros, so will be more predicatable testing for the
stable trees in the future.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
Reviewed-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
... in the style of debian:12-ppc64le.
Rename the jobs and reposition them later as they're not a dependency for the
smoke testing any more.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
qemu-system-ppc64/8.1.0-ppc64 was added because bullseye's QEMU didn't
understand the powernv9 machine. However bookworm's QEMU does and this is
preferable to maintaining a random build of QEMU ourselves.
Use the debian:12-ppc64le container and test the output of that build too.
Remove qemu-system-ppc64-8.1.0-ppc64-export which is unused now.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
... conforming to the new naming scheme; $DISTRO-$VERSION-$ARCH-* so the jobs
sort more coherently.
Make it non-root by default, and set XEN_TARGET_ARCH=ppc64. Include QEMU too,
which will be used subsequently.
Add build jobs too, with debian-12-ppc64le-gcc-debug specifically early as it
will be used for smoke testing shortly.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Archlinux is a rolling distro. As a consequence, rebuilding the container
periodically changes the toolchain, and this affects all stable branches in
one go.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Randconfig builds pick CONFIG_DEBUG with 50% probability. Therefore
$foo{,-debug}-randconfig are just two equivalent jobs with misleading names.
Furthermore, arm64 has randconfig jobs for both Alpine and Bookworm; Both
environments use GCC 12.2, so we have 4 equivalent jobs. Delete 3 of them.
This leaves us with 4 primary randconfig jobs:
gitlab-ci/build.yaml:381:debian-bookworm-gcc-arm32-randconfig:
gitlab-ci/build.yaml:429:alpine-3.18-gcc-arm64-randconfig:
gitlab-ci/build.yaml:495:archlinux-current-gcc-riscv64-randconfig:
gitlab-ci/build.yaml:640:debian-bookworm-gcc-randconfig:
as PPC64 doesn't want randconfig right now, and buster-gcc-ibt is a special
job with a custom compiler.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Right now, most build containers use root. Archlinux, Fedora and Yocto set up
a regular user called `user`.
For those containers, trying to containerize as root fails, because
CONTAINER_UID0=1 does nothing, whereas CONTAINER_UID0=0 forces the user away
from root.
To make CONTAINER_UID0=1 work reliably, force to root if requested.
Fixes: 17fbe6504d ("automation: introduce a new variable to control container user")
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
The history on this one is complicated. The note to README was added in
commit 1f95747a4f ("Add openssl-dev and xorg-x11-dev to README") in 2007.
At the time, there was a vendered version of Qemu in xen.git with a local
modification using <X11/keysymdef.h> to access the monitor console over VNC.
The final reference to keysymdef.h was dropped in commit 85896a7c4d ("build:
add autoconf to replace custom checks in tools/check") in 2012. The next
prior mention was in 2009 with commit a8ccb671c3 ("tools: fix x11 check")
noting that x11 was not a direct dependcy of Xen; it was transitive through
SDL for Qemu for source-based distros.
It appears there may have been other unspecified dependencies on xorg,
e.g. the use of lndir by unmodified_drivers which are no longer relevant
either.
These days its only the Debian based dockerfiles which install xorg-x11, and
Qemu builds fine in these and others without x11.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Roger Pau Monné <roger.pau@citrix.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
The final user was blktap2, removed in Xen 4.13.
Strip out the ./configure check for it, and stop explicitly installing it in
the FreeBSD testing.
Fixes: 2a556b63a2 ("Drop blktap2")
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Roger Pau Monné <roger.pau@citrix.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
set_cloexec() had a bit too much copy&pate from setnonblock(), and
insufficient testing on ancient versions of Linux...
As written (emulating ancient linux by undef'ing O_CLOEXEC), strace shows:
open("/dev/xen/xenbus", O_RDWR) = 3
fcntl(3, F_GETFL) = 0x8002 (flags O_RDWR|O_LARGEFILE)
fcntl(3, 0x8003 /* F_??? */, 0x7ffe4a771d90) = -1 EINVAL (Invalid argument)
close(3) = 0
which is obviously nonsense.
Switch F_GETFL -> F_GETFD, and fix the second invocation to use F_SETFD. With
this, strace is rather happer:
open("/dev/xen/xenbus", O_RDWR) = 3
fcntl(3, F_GETFD) = 0
fcntl(3, F_SETFD, FD_CLOEXEC) = 0
Fixes: bf7c146470 ("tools/libxs: Fix CLOEXEC handling in get_dev()")
Reported-by: Ross Lagerwall <ross.lagerwall@citrix.com>
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Ross Lagerwall <ross.lagerwall@citrix.com>
Reviewed-by: Juergen Gross <jgross@suse.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Update to Ubuntu 24.04, and checkout@v4 as v2 is deprecated.
The build step goes out of it's way to exclude docs and stubdom (but include
plain MiniOS), so disable those at the ./configure stage.
Refresh the package list. libbz2-dev was in there twice, and e2fslibs-dev is
a a transitional package to libext2fs-dev. I'm not aware of libtool ever
having been a Xen dependency.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Roger Pau Monné <roger.pau@citrix.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
We previously upped the minimum python version to 2.7, but neglected to
reflect this in ./configure
Fixes: 2a353c048c ("tools: Don't use distutils in configure or Makefile")
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Roger Pau Monné <roger.pau@citrix.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Fix bug introduced by 0762e2502f ("x86/physdev: factor out the code to allocate and
map a pirq"). After that re-factoring, when pirq<0 and current_pirq>0, it means
caller want to allocate a free pirq for irq but irq already has a mapped pirq, then
it returns the negative pirq, so it fails. However, the logic before that
re-factoring is different, it should return the current_pirq that irq was already
mapped to and make the call success.
Fixes: 0762e2502f ("x86/physdev: factor out the code to allocate and map a pirq")
Signed-off-by: Jiqian Chen <Jiqian.Chen@amd.com>
Signed-off-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Jiqian Chen <Jiqian.Chen@amd.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
CentOS 7 is fully End-of-life as of 2024-06-30, and the Yum repo configuration
points at URLs which have become non-existent.
First, start by using a heredoc RUN for legibility. It's important to use
`set -e` to offset the fact that we're no longer chaining every command
together with an &&.
Also, because we're using a single RUN command to perform all RPM operations,
we no longer need to work around the OverlayFS bug.
Adjust the CentOS-*.repo files to point at vault.centos.org. This also
involves swapping mirrorlist= for baseurl= in the yum config.
Use a minor bashism to express the dependenices more coherently, and identify
why we have certain dependencies. Some adjustments are:
* We need bzip2-devel for the dombuilder. bzip2 needs retaining stubdom or
`tar` fails to unpack the .bz2 archives.
* {lzo,lz4,ztd}-devel are new optional dependency since the last time this
package list was refreshed.
* openssl-devel hasn't been a dependency since Xen 4.6.
* We long ago ceased being able to build Qemu and SeaBIOS in this container,
so drop their dependencies too.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Roger Pau Monné <roger.pau@citrix.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Commit b5739330d7 introduced the use of heredocs in the jessie/stretch
dockerfiles.
It turns out this was introduced by BuildKit in 2018 along with a
standardisation of Dockerfile syntax, and has subsequently been adopted by the
docker community.
Annotate all dockerfiles with a statement of the syntax in use, and extend
README.md details including how to activate BuildKit when it's available but
off by default.
This allows the containers to be rebuilt following commit a0e29b3163 ("CI:
Drop glibc-i386 from the build containers").
Fixes: b5739330d7 ("automation: fix jessie/stretch images to use archive.debian.org apt repos")
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Roger Pau Monné <roger.pau@citrix.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
This was clearly written by a C programmer, rather than a python programmer.
Drop all the useless semi-colons.
The very final line of the script simply references f.close, rather than
calling the function. Switch to using a with: statement, as python does care
about unclosed files if you enable enough warnings.
No functional change.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Prior to the split, "the clients" used tools/xenstored/Makefile.common whose
clean rule includes *.o whereas after the split, the removal of *.o was lost
by virtule of not including Makefile.common any more.
This is the bug behind the following build error:
make[2]: Entering directory '/local/xen.git/tools/xs-clients'
gcc xenstore_client.o (snip)
/usr/bin/ld: xenstore_client.o: relocation R_X86_64_32S against `.rodata' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: failed to set dynamic section sizes: bad value
collect2: error: ld returned 1 exit status
make[2]: *** [Makefile:35: xenstore] Error 1
which was caused by `make clean` not properly cleaning the tree as I was
swapping between various build containers.
Switch to a plain single-colon clean rule.
Fixes: 5c293058b1 ("tools/xenstore: move xenstored sources into dedicated directory")
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Juergen Gross <jgross@suse.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
In it's current form it's rather cryptic for uninitiated, so this patch
documents everything in there to make it more palatable.
No functional change.
Signed-off-by: Alejandro Vallejo <alejandro.vallejo@cloud.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Support for specifying "raw" insns was added only in 2.38.
To support older version it would be better switch to .insn
with operands.
The following compilation error occurs:
./arch/riscv/include/asm/processor.h: Assembler messages:
./arch/riscv/include/asm/processor.h:70: Error: unrecognized opcode `0x0100000F'
In case of the following Binutils:
$ riscv64-linux-gnu-as --version
GNU assembler (GNU Binutils for Debian) 2.35.2
Suggested-by: Andrew Cooper <andrew.cooper3@citrix.com>
Suggested-by: Jan Beulich <jbeulich@suse.com>
Signed-off-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
PVH in particular has no (externally visible) notion of pIRQ-s. Mention
that in the description of the respective command line option and have
arch_hwdom_irqs() also reflect this (thus suppressing the log message
there as well, as being pretty meaningless in this case anyway).
Suggested-by: Roger Pau Monné <roger.pau@citrix.com>
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Roger Pau Monné <roger.pau@citrix.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Much like noted in 43d5c5d5f7 ("xen: avoid UB in guest handle
arithmetic"), address calculations involved in accessing a struct field
can overflow, too. Cast respective pointers to "unsigned long" and
convert type checking accordingly. Remaining arithmetic is, despite
there possibly being mathematical overflow, okay as per the C99 spec:
"A computation involving unsigned operands can never overflow, because a
result that cannot be represented by the resulting unsigned integer type
is reduced modulo the number that is one greater than the largest value
that can be represented by the resulting type." The overflow that we
need to guard against is checked for in array_access_ok().
While there add the missing (see {,__}copy_to_guest_offset()) is-not-
const checks to {,__}copy_field_to_guest().
Typically, but not always, no change to generated code; code generation
(register allocation) is different for at least common/grant_table.c.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
While doing so is intentional when invoking the actual callback, to
mimic a hard-coded SYCALL_MASK / FMASK MSR, the same should not be done
when no handler is available and hence #UD is raised.
Fixes: ca6fcf4321 ("x86/pv: Inject #UD for missing SYSCALL callbacks")
Reported-by: Andrew Cooper <andrew.cooper3@citrix.com>
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
PHYSDEVOP_pirq_eoi_gmfn_v<N> accepting just a single GFN implies that no
more than 32k pIRQ-s can be used by a domain on x86. Document this upper
bound.
To also enforce the limit, (ab)use both arch_hwdom_irqs() (changing its
parameter type) and setup_system_domains(). This is primarily to avoid
exposing the two static variables or introducing yet further arch hooks.
While touching arch_hwdom_irqs() also mark it hwdom-init.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Roger Pau Monné <roger.pau@citrix.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
xs_fileno() opens a pipe on first use to communicate between the watch thread
and the main thread. Nothing ever sets CLOEXEC on the file descriptors.
Check for the availability of the pipe2() function with configure. Despite
starting life as Linux-only, FreeBSD and NetBSD have gained it.
When pipe2() isn't available, try our best with pipe() and set_cloexec().
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Juergen Gross <jgross@suse.com>
Acked-by: Anthony PERARD <anthony.perard@vates.tech>
get_socket() opens a socket, then uses fcntl() to set CLOEXEC. This is racy
with exec().
Open the socket with SOCK_CLOEXEC. Use the same compatibility strategy as
O_CLOEXEC on ancient versions of Linux.
Reported-by: Frediano Ziglio <frediano.ziglio@cloud.com>
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Juergen Gross <jgross@suse.com>
Acked-by: Anthony PERARD <anthony.perard@vates.tech>
Move the O_CLOEXEC compatibility outside of an #ifdef USE_PTHREAD block.
Introduce set_cloexec() to wrap fcntl() setting FD_CLOEXEC. It will be reused
for other CLOEXEC fixes too.
Use set_cloexec() when O_CLOEXEC isn't available as a best-effort fallback.
Fixes: f4f2f3402b ("tools/libxs: Open /dev/xen/xenbus fds as O_CLOEXEC")
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Juergen Gross <jgross@suse.com>
Acked-by: Anthony PERARD <anthony.perard@vates.tech>
xc_dom_alloc_segment() is passed a size in bytes, calculates a size in pages
from it, then fills in the new segment information with a bytes value
re-calculated from the number of pages.
This causes the module information given to the guest (MB, or PVH) to have
incorrect sizes; specifically, sizes rounded up to the next page.
This in turn is problematic for Xen. When Xen finds a gzipped module, it
peeks at the end metadata to judge the decompressed size, which is a -4
backreference from the reported end of the module.
Fill in seg->vend using the correct number of bytes.
Fixes: ea7c8a3d0e ("libxc: reorganize domain builder guest memory allocator")
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Anthony PERARD <anthony.perard@vates.tech>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Julien Grall