Paolo Bonzini [Tue, 20 Nov 2018 18:41:48 +0000 (19:41 +0100)]
nvme: fix out-of-bounds access to the CMB
Because the CMB BAR has a min_access_size of 2, if you read the last
byte it will try to memcpy *2* bytes from n->cmbuf, causing an off-by-one
error. This is CVE-2018-16847.
Another way to fix this might be to register the CMB as a RAM memory
region, which would also be more efficient. However, that might be a
change for big-endian machines; I didn't think this through and I don't
know how real hardware works. Add a basic testcase for the CMB in case
somebody does this change later on.
Igor Druzhinin [Tue, 6 Nov 2018 12:16:55 +0000 (12:16 +0000)]
nvme: call blk_drain in NVMe reset code to avoid lockups
When blk_flush called in NVMe reset path S/C queues are already freed
which means that re-entering AIO handling loop having some IO requests
unfinished will lockup or crash as their SG structures being potentially
reused. Call blk_drain before freeing the queues to avoid this nasty
scenario.
iotests: fix nbd test 233 to work correctly with raw images
The first qemu-io command must honour the $IMGFMT that is set rather
than hardcoding qcow2. The qemu-nbd commands should also set $IMGFMT
to avoid the insecure format probe warning.
Alberto Garcia [Mon, 12 Nov 2018 14:00:48 +0000 (16:00 +0200)]
block: Fix update of BDRV_O_AUTO_RDONLY in update_flags_from_options()
Commit e35bdc123a4ace9f4d3fcca added the auto-read-only option and the
code to update its corresponding flag in update_flags_from_options(),
but forgot to clear the flag if auto-read-only is false.
scsi-disk: Fix crash if underlying host file or disk returns error
Commit 40dce4ee6 "scsi-disk: fix rerror/werror=ignore" introduced a
bug which causes qemu to crash with the assertion error below if the
host file or disk returns an error:
< kwolf> Hm, should the final return false; in that patch
actually be a return true?
< kwolf> Because I think he didn't intend to change anything
except BLOCK_ERROR_ACTION_IGNORE
Eric Blake [Tue, 20 Nov 2018 22:52:41 +0000 (16:52 -0600)]
iotests: Skip 233 if certtool not installed
The use of TLS while building qemu is optional. While the
'certtool' binary should be available on every platform that
supports building against TLS, that does not imply that the
developer has installed it. Make the test gracefully skip
in that case.
Kevin Wolf [Tue, 20 Nov 2018 17:12:21 +0000 (18:12 +0100)]
iotests: Replace assertEquals() with assertEqual()
TestCase.assertEquals() is deprecated since Python 2.7. Recent Python
versions print a warning when the function is called, which makes test
cases fail.
Replace it with the preferred spelling assertEqual().
Kevin Wolf [Tue, 20 Nov 2018 17:09:49 +0000 (18:09 +0100)]
iotests: Replace time.clock() with Timeout
time.clock() is deprecated since Python 3.3. Current Python versions
warn that the function will be removed in Python 3.8, and those warnings
make the test case 118 fail.
Replace it with the Timeout mechanism that is compatible with both
Python 2 and 3, and makes the code even a little nicer.
Greg Kurz [Tue, 20 Nov 2018 12:00:35 +0000 (13:00 +0100)]
9p: take write lock on fid path updates (CVE-2018-19364)
Recent commit 5b76ef50f62079a fixed a race where v9fs_co_open2() could
possibly overwrite a fid path with v9fs_path_copy() while it is being
accessed by some other thread, ie, use-after-free that can be detected
by ASAN with a custom 9p client.
It turns out that the same can happen at several locations where
v9fs_path_copy() is used to set the fid path. The fix is again to
take the write lock.
* remotes/huth-gitlab/tags/pull-request-2018-11-20:
qtest: log QEMU command line
tests/ide: Free pcibus when finishing a test
tests: add qmp/missing-any-arg test
Thomas Huth [Tue, 13 Nov 2018 15:03:21 +0000 (16:03 +0100)]
tests/ide: Free pcibus when finishing a test
Once a test has finished, the pcibus structure should be freed, to
avoid leaking memory and to make sure that the structure is properly
re-initialized when the next test starts.
test_qmp_missing_any_arg() is about a bug in infrastructure used by
the QMP core, fixed in commit c489780203. We covered the bug in
infrastructure unit tests (commit bce3035a44). Let's test
it at the QMP level as well.
Peter Maydell [Mon, 19 Nov 2018 18:11:10 +0000 (18:11 +0000)]
Merge remote-tracking branch 'remotes/ericb/tags/pull-nbd-2018-11-19' into staging
nbd patches for 2018-11-19
Add iotest coverage for NBD connections using TLS, including
a couple of code fixes that it pointed out
- Mao Zhongyi: 0/3 Do some cleaning work in qemu-iotests
- Daniel P. Berrangé: io: return 0 for EOF in TLS session read after shutdown
- Daniel P. Berrangé: 0/6 Misc fixes to NBD
- Eric Blake: iotests: Drop use of bash keyword 'function'
# gpg: Signature made Mon 19 Nov 2018 17:43:32 GMT
# gpg: using RSA key A7A16B4A2527436A
# gpg: Good signature from "Eric Blake <[email protected]>"
# gpg: aka "Eric Blake (Free Software Programmer) <[email protected]>"
# gpg: aka "[jpeg image of size 6874]"
# Primary key fingerprint: 71C2 CC22 B1C4 6029 27D2 F3AA A7A1 6B4A 2527 436A
* remotes/ericb/tags/pull-nbd-2018-11-19:
iotests: Drop use of bash keyword 'function'
iotests: Also test I/O over NBD TLS
tests: exercise NBD server in TLS mode
tests: add iotests helpers for dealing with TLS certificates
tests: check if qemu-nbd is still alive before waiting
tests: pull qemu-nbd iotest helpers into common.nbd file
io: return 0 for EOF in TLS session read after shutdown
nbd/server: Ignore write errors when replying to NBD_OPT_ABORT
nbd: fix whitespace in server error message
qemu-iotests: Modern shell scripting (use $() instead of ``)
qemu-iotests: convert `pwd` and $(pwd) to $PWD
qemu-iotests: remove unused variable 'here'
Eric Blake [Fri, 16 Nov 2018 21:50:02 +0000 (15:50 -0600)]
iotests: Drop use of bash keyword 'function'
Bash allows functions to be declared with or without the leading
keyword 'function'; but including the keyword does not comply with
POSIX syntax, and is confusing to ksh users where the use of the
keyword changes the scoping rules for functions. Stick to the
POSIX form through iotests.
Done mechanically with:
sed -i 's/^function //' $(git ls-files tests/qemu-iotests)
tests: check if qemu-nbd is still alive before waiting
If the qemu-nbd UNIX socket has not shown up, the tests will sleep a bit
and then check again repeatedly for up to 30 seconds. This is pointless
if the qemu-nbd process has quit due to an error, so check whether the
pid is still alive before waiting and retrying.
io: return 0 for EOF in TLS session read after shutdown
GNUTLS takes a paranoid approach when seeing 0 bytes returned by the
underlying OS read() function. It will consider this an error and
return GNUTLS_E_PREMATURE_TERMINATION instead of propagating the 0
return value. It expects apps to arrange for clean termination at
the protocol level and not rely on seeing EOF from a read call to
detect shutdown. This is to harden apps against a malicious 3rd party
causing termination of the sockets layer.
This is unhelpful for the QEMU NBD code which does have a clean
protocol level shutdown, but still relies on seeing 0 from the I/O
channel read in the coroutine handling incoming replies.
The upshot is that when using a plain NBD connection shutdown is
silent, but when using TLS, the client spams the console with
Cannot read from TLS channel: Broken pipe
The NBD connection has, however, called qio_channel_shutdown()
at this point to indicate that it is done with I/O. This gives
the opportunity to optimize the code such that when the channel
has been shutdown in the read direction, the error code
GNUTLS_E_PREMATURE_TERMINATION gets turned into a '0' return
instead of an error.
Eric Blake [Sat, 17 Nov 2018 22:32:21 +0000 (16:32 -0600)]
nbd/server: Ignore write errors when replying to NBD_OPT_ABORT
Commit 37ec36f6 intentionally ignores errors when trying to reply
to an NBD_OPT_ABORT request for plaintext clients, but did not make
the same change for a TLS server. Since NBD_OPT_ABORT is
documented as being a potential for an EPIPE when the client hangs
up without waiting for our reply, we don't need to pollute the
server's output with that failure.
Mao Zhongyi [Wed, 24 Oct 2018 09:40:51 +0000 (17:40 +0800)]
qemu-iotests: Modern shell scripting (use $() instead of ``)
Various shell files contain a mix between obsolete ``
and modern $(); It would be nice to convert to using
$() everywhere. For now, just do the qemu-iotests directory.
Mao Zhongyi [Fri, 16 Nov 2018 20:33:03 +0000 (14:33 -0600)]
qemu-iotests: convert `pwd` and $(pwd) to $PWD
POSIX requires $PWD to be reliable, and we expect all
shells used by qemu scripts to be relatively close to
POSIX. Thus, it is smarter to avoid forking the pwd
executable for something that is already available in
the environment.
So replace it with the following:
sed -i 's/\(`pwd`\|\$(pwd)\)/$PWD/g' $(git grep -l pwd)
Then delete a pointless line assigning PWD to itself.
Mao Zhongyi [Wed, 24 Oct 2018 09:40:50 +0000 (17:40 +0800)]
qemu-iotests: remove unused variable 'here'
Running
git grep '\$here' tests/qemu-iotests
has 0 hits, which means we are setting a variable that has
no use. It appears that commit e8f8624d removed the last
use. So execute the following cmd to remove all of
the 'here=...' lines as dead code.
sed -i '/^here=/d' $(git grep -l '^here=' tests/qemu-iotests)
Peter Maydell [Mon, 19 Nov 2018 15:58:17 +0000 (15:58 +0000)]
Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20181119' into staging
target-arm queue:
* various MAINTAINERS file updates
* hw/block/onenand: use qemu_log_mask() for reporting
* hw/block/onenand: Fix off-by-one error allowing out-of-bounds read
on the n800 and n810 machine models
* target/arm: fix smc incorrectly trapping to EL3 when secure is off
* hw/arm/stm32f205: Fix the UART and Timer region size
* target/arm: read ID registers for KVM guests so they can be
used to gate "is feature X present" checks
* remotes/pmaydell/tags/pull-target-arm-20181119:
MAINTAINERS: list myself as maintainer for various Arm boards
hw/block/onenand: use qemu_log_mask() for reporting
hw/block/onenand: Fix off-by-one error allowing out-of-bounds read
target/arm: fix smc incorrectly trapping to EL3 when secure is off
hw/arm/stm32f205: Fix the UART and Timer region size
MAINTAINERS: Add entries for missing ARM boards
target/arm: Fill in ARMISARegisters for kvm32
target/arm: Introduce read_sys_reg32 for kvm32
target/arm: Fill in ARMISARegisters for kvm64
target/arm: Install ARMISARegisters from kvm host
Peter Maydell [Mon, 19 Nov 2018 15:29:08 +0000 (15:29 +0000)]
MAINTAINERS: list myself as maintainer for various Arm boards
In practice for most of the more-or-less orphan Arm board models,
I will review patches and put them in via the target-arm tree.
So list myself as an "Odd Fixes" status maintainer for them.
This commit downgrades these boards to "Odd Fixes":
* Allwinner-A10
* Exynos
* Calxeda Highbank
* Canon DIGIC
* Musicpal
* nSeries
* Palm
* PXA2xx
These boards were already "Odd Fixes":
* Gumstix
* i.MX31 (kzm)
Philippe Mathieu-Daudé has requested to be moved to R:
status for Gumstix now that I am listed as the M: contact.
Some boards are maintained, but their patches still go
via the target-arm tree, so add myself as a secondary
maintainer contact for those:
* Xilinx Zynq
* Xilinx ZynqMP
* STM32F205
* Netduino 2
* SmartFusion2
* Mecraft M2S-FG484
* ASPEED BMCs
* NRF51
Peter Maydell [Mon, 19 Nov 2018 15:29:08 +0000 (15:29 +0000)]
hw/block/onenand: use qemu_log_mask() for reporting
Update the onenand device to use qemu_log_mask() for reporting
guest errors and unimplemented features, rather than plain
fprintf() and hw_error().
(We leave the hw_error() in onenand_reset(), as that is
triggered by a failure to read the underlying block device
for the bootRAM, not by guest action.)
Luc Michel [Mon, 19 Nov 2018 15:29:08 +0000 (15:29 +0000)]
target/arm: fix smc incorrectly trapping to EL3 when secure is off
This commit fixes a case where the CPU would try to go to EL3 when
executing an smc instruction, even though ARM_FEATURE_EL3 is false. This
case is raised when the PSCI conduit is set to smc, but the smc
instruction does not lead to a valid PSCI call.
QEMU crashes with an assertion failure latter on because of incoherent
mmu_idx.
This commit refactors the pre_smc helper by enumerating all the possible
way of handling an scm instruction, and covering the previously missing
case leading to the crash.
The following minimal test would crash before this commit:
Seth Kintigh [Mon, 19 Nov 2018 15:29:08 +0000 (15:29 +0000)]
hw/arm/stm32f205: Fix the UART and Timer region size
The UART and timer devices for the stm32f205 were being created
with memory regions that were too large. Use the size specified
in the chip datasheet.
The old sizes were so large that the devices would overlap with
each other in the SoC memory map, so this fixes a bug that
caused odd behavior and/or crashes when trying to set up multiple
UARTs.
Thomas Huth [Mon, 19 Nov 2018 15:29:08 +0000 (15:29 +0000)]
MAINTAINERS: Add entries for missing ARM boards
Add entries for the boards "mcimx6ul-evk", "mcimx7d-sabre", "raspi2",
"raspi3", "sabrelite", "vexpress-a15", "vexpress-a9" and "virt".
While we're at it, also adjust the "i.MX31" section a little bit,
so that the wildcards there do not match anymore for unrelated files
(e.g. the new hw/misc/imx6ul_ccm.c file).
Peter Maydell [Mon, 19 Nov 2018 14:31:48 +0000 (14:31 +0000)]
Merge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging
Block layer patches:
- file-posix: Fix shared permission locks after reopen
- block: Fix error path for failed .bdrv_reopen_prepare
- qcow2: Catch invalid allocations when the image becomes too large
- vvfat/fdc/nvme: Fix segfaults and leaks
# gpg: Signature made Mon 19 Nov 2018 14:28:18 GMT
# gpg: using RSA key 7F09B272C88F2FD6
# gpg: Good signature from "Kevin Wolf <[email protected]>"
# Primary key fingerprint: DC3D EB15 9A9A F95D 3D74 56FE 7F09 B272 C88F 2FD6
* remotes/kevin/tags/for-upstream:
iotests: Test file-posix locking and reopen
file-posix: Fix shared locks on reopen commit
block: Always abort reopen after prepare succeeded
iotests: Add new test 220 for max compressed cluster offset
qcow2: Don't allow overflow during cluster allocation
qcow2: Document some maximum size constraints
vvfat: Fix memory leak
fdc: fix segfault in fdctrl_stop_transfer() when DMA is disabled
nvme: fix oob access issue(CVE-2018-16847)
Max Reitz [Fri, 16 Nov 2018 16:45:25 +0000 (17:45 +0100)]
file-posix: Fix shared locks on reopen commit
s->locked_shared_perm is the set of bits locked in the file, which is
the inverse of the permissions actually shared. So we need to pass them
as they are to raw_apply_lock_bytes() instead of inverting them again.
Max Reitz [Fri, 16 Nov 2018 16:45:24 +0000 (17:45 +0100)]
block: Always abort reopen after prepare succeeded
bdrv_reopen_multiple() does not invoke bdrv_reopen_abort() for the
element of the reopen queue for which bdrv_reopen_prepare() failed,
because it assumes that the prepare function will have rolled back all
changes already.
However, bdrv_reopen_prepare() does not do this in every case: It may
notice an error after BlockDriver.bdrv_reopen_prepare() succeeded, and
it will not invoke BlockDriver.bdrv_reopen_abort() then; and neither
will bdrv_reopen_multiple(), as explained above.
This is wrong because we must always call .bdrv_reopen_commit() or
.bdrv_reopen_abort() after .bdrv_reopen_prepare() has succeeded.
Otherwise, the block driver has no chance to undo what it has done in
its implementation of .bdrv_reopen_prepare().
To fix this, bdrv_reopen_prepare() has to call .bdrv_reopen_abort() if
it wants to return an error after .bdrv_reopen_prepare() has succeeded.
Eric Blake [Tue, 13 Nov 2018 23:03:19 +0000 (17:03 -0600)]
iotests: Add new test 220 for max compressed cluster offset
If you have a capable file system (tmpfs is good, ext4 not so much;
run ./check with TEST_DIR pointing to a good location so as not
to skip the test), it's actually possible to create a qcow2 file
that expands to a sparse 512T image with just over 38M of content.
The test is not the world's fastest (qemu crawling through 256M
bits of refcount table to find the next cluster to allocate takes
several seconds, as does qemu-img check reporting millions of
leaked clusters); but it DOES catch the problem that the previous
patch just fixed where writing a compressed cluster to a full
image ended up overwriting the wrong cluster.
Eric Blake [Tue, 13 Nov 2018 23:03:18 +0000 (17:03 -0600)]
qcow2: Don't allow overflow during cluster allocation
Our code was already checking that we did not attempt to
allocate more clusters than what would fit in an INT64 (the
physical maximimum if we can access a full off_t's worth of
data). But this does not catch smaller limits enforced by
various spots in the qcow2 image description: L1 and normal
clusters of L2 are documented as having bits 63-56 reserved
for other purposes, capping our maximum offset at 64PB (bit
55 is the maximum bit set). And for compressed images with
2M clusters, the cap drops the maximum offset to bit 48, or
a maximum offset of 512TB. If we overflow that offset, we
would write compressed data into one place, but try to
decompress from another, which won't work.
It's actually possible to prove that overflow can cause image
corruption without this patch; I'll add the iotests separately
in the next commit.
Eric Blake [Thu, 15 Nov 2018 18:34:08 +0000 (12:34 -0600)]
qcow2: Document some maximum size constraints
Although off_t permits up to 63 bits (8EB) of file offsets, in
practice, we're going to hit other limits first. Document some
of those limits in the qcow2 spec (some are inherent, others are
implementation choices of qemu), and how choice of cluster size
can influence some of the limits.
While we cannot map any uncompressed virtual cluster to any
address higher than 64 PB (56 bits) (due to the current L1/L2
field encoding stopping at bit 55), qemu's cap of 8M for the
refcount table can still access larger host addresses for some
combinations of large clusters and small refcount_order. For
comparison, ext4 with 4k blocks caps files at 16PB.
Another interesting limit: for compressed clusters, the L2 layout
requires an ever-smaller maximum host offset as cluster size gets
larger, down to a 512 TB maximum with 2M clusters. In particular,
note that with a cluster size of 8k or smaller, the L2 entry for
a compressed cluster could technically point beyond the 64PB mark,
but when you consider that with 8k clusters and refcount_order = 0,
you cannot access beyond 512T without exceeding qemu's limit of an
8M cap on the refcount table, it is unlikely that any image in the
wild has attempted to do so. To be safe, let's document that bits
beyond 55 in a compressed cluster must be 0.
Mark Cave-Ayland [Sun, 11 Nov 2018 09:40:23 +0000 (09:40 +0000)]
fdc: fix segfault in fdctrl_stop_transfer() when DMA is disabled
Commit c8a35f1cf0f "fdc: use IsaDma interface instead of global DMA_*
functions" accidentally introduced a segfault in fdctrl_stop_transfer() for
non-DMA transfers.
If fdctrl->dma_chann has not been configured then the fdctrl->dma interface
reference isn't initialised during isabus_fdc_realize(). Unfortunately
fdctrl_stop_transfer() unconditionally references the DMA interface when
finishing the transfer causing a NULL pointer dereference.
Fix the issue by adding a check in fdctrl_stop_transfer() so that the DMA
interface reference and release method is only invoked if fdctrl->dma_chann
has been set.
(This issue was discovered by Martin testing a recent change in the NetBSD
installer under qemu-system-sparc)
Li Qiang [Fri, 2 Nov 2018 01:22:43 +0000 (18:22 -0700)]
nvme: fix oob access issue(CVE-2018-16847)
Currently, the nvme_cmb_ops mr doesn't check the addr and size.
This can lead an oob access issue. This is triggerable in the guest.
Add check to avoid this issue.
Cornelia Huck [Fri, 26 Oct 2018 10:57:11 +0000 (12:57 +0200)]
MAINTAINERS: clarify some of the tags
The MAINTAINERS file is a bit sparse on information about what
the different designators are. Let's add some more information
to give contributors a better idea about what the different
roles are.
Peter Maydell [Mon, 19 Nov 2018 10:23:45 +0000 (10:23 +0000)]
Merge remote-tracking branch 'remotes/amarkovic/tags/mips-queue-november-2018-v2' into staging
MIPS queue for QEMU 3.1-rc2 - v2
# gpg: Signature made Sat 17 Nov 2018 18:30:46 GMT
# gpg: using RSA key D4972A8967F75A65
# gpg: Good signature from "Aleksandar Markovic <[email protected]>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 8526 FBF1 5DA3 811F 4A01 DD75 D497 2A89 67F7 5A65
* remotes/amarkovic/tags/mips-queue-november-2018-v2:
MAINTAINERS: Add Stefan Markovic as a MIPS reviewer
target/mips: Disable R5900 support
target/mips: Rename MMI-related functions
target/mips: Rename MMI-related opcodes
target/mips: Rename MMI-related masks
target/mips: Guard check_insn with INSN_R5900 check
target/mips: Guard check_insn_opc_user_only with INSN_R5900 check
target/mips: Fix decoding mechanism of special R5900 opcodes
target/mips: Fix decoding mechanism of R5900 DIV1 and DIVU1
target/mips: Fix decoding mechanism of R5900 MFLO1, MFHI1, MTLO1 and MTHI1
linux-user: Update MIPS specific prctl() implementation
Peter Maydell [Mon, 19 Nov 2018 09:35:27 +0000 (09:35 +0000)]
Merge remote-tracking branch 'remotes/riscv/tags/riscv-for-master-3.1-rc2' into staging
RISC-V Patches for 3.1-rc2
This pull request contains four patches that aren't really related to
each other aside from all being bug fixes that I think should go in for
3.1.0:
* The second half of Alistair's memory leak patch set that I missed last
week.
* A fix to make fclass.d availiable only on RV64IFD systems (without
this it's availiable on RV32IFD systems, truncating the result).
* A fix to make sfence.vm availiable only in priv-1.9.1, and sfence.vma
only availiable in priv-1.10.
* A change to respect fences in user-mode emulators, which were
previously treated as NOPs.
As usual, this builds and boot Linux for me. I don't think I have
anything else planned for 3.1.0, but I may be wrong as things are a bit
hectic this week.
# gpg: Signature made Tue 13 Nov 2018 23:48:38 GMT
# gpg: using RSA key EF4CA1502CCBAB41
# gpg: Good signature from "Palmer Dabbelt <[email protected]>"
# gpg: aka "Palmer Dabbelt <[email protected]>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 00CE 76D1 8349 60DF CE88 6DF8 EF4C A150 2CCB AB41
* remotes/riscv/tags/riscv-for-master-3.1-rc2:
RISC-V: Respect fences for user-only emulators
target/riscv: Fix sfence.vm/a both available in any priv version
target/riscv: Fix FCLASS_D being treated as RV64 only
hw/riscv/virt: Free the test device tree node name
MAINTAINERS: Add Stefan Markovic as a MIPS reviewer
Add Stefan Markovic as a MIPS reviewer. He had several key
contributions to QEMU for MIPS this year. He is a meticulous
person with the ability to think and act on many levels.
Disable R5900 support. There are some outstanding issues related
to ABI support and emulation accuracy, that were not understood
well during review process. Disable to avoid backward compatibility
issues.
Fredrik Noring [Wed, 7 Nov 2018 19:19:14 +0000 (20:19 +0100)]
target/mips: Fix decoding mechanism of special R5900 opcodes
MOVN, MOVZ, MFHI, MFLO, MTHI, MTLO, MULT, MULTU, DIV, DIVU, DMULT,
DMULTU, DDIV, DDIVU and JR are decoded in decode_opc_special_tx79
instead of the generic decode_opc_special_legacy.
* remotes/kraxel/tags/fixes-31-20181116-pull-request:
help: Provide help for egl-headless
ui: Allow specifying 'rendernode' display option for egl-headless
qapi: Add "rendernode" display option for egl-headless
Erik Skultety [Fri, 16 Nov 2018 10:14:44 +0000 (11:14 +0100)]
help: Provide help for egl-headless
EGL headless has been missing from QEMU's help or man page, we should
mention that such a thing exists, especially since projects like libvirt
might rely on that. This patch also adds the newly introduced option for
egl-headless 'rendernode'.
Erik Skultety [Fri, 16 Nov 2018 10:14:43 +0000 (11:14 +0100)]
ui: Allow specifying 'rendernode' display option for egl-headless
As libvirt can't predict which rendernode QEMU would pick, it
won't adjust the permissions on the device, hence QEMU getting
"Permission denied" when opening the DRI device. Therefore, enable
'rendernode' option for egl-headless display type.
Erik Skultety [Fri, 16 Nov 2018 10:14:42 +0000 (11:14 +0100)]
qapi: Add "rendernode" display option for egl-headless
Unlike SPICE, egl-headless doesn't offer a way of specifying the DRM
node used for OpenGL, hence QEMU always selecting the first one that is
available. Thus, add the 'rendernode' option for egl-headless to QAPI.
Peter Maydell [Fri, 16 Nov 2018 10:01:04 +0000 (10:01 +0000)]
Merge remote-tracking branch 'remotes/jasowang/tags/net-pull-request' into staging
# gpg: Signature made Fri 16 Nov 2018 03:38:58 GMT
# gpg: using RSA key EF04965B398D6211
# gpg: Good signature from "Jason Wang (Jason Wang on RedHat) <[email protected]>"
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg: It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 215D 46F4 8246 689E C77F 3562 EF04 965B 398D 6211
Peter Maydell [Thu, 15 Nov 2018 14:18:25 +0000 (14:18 +0000)]
Merge remote-tracking branch 'remotes/stefanberger/tags/pull-tpm-2018-11-15-1' into staging
Merge tpm 2018/11/15 v1
# gpg: Signature made Thu 15 Nov 2018 14:03:45 GMT
# gpg: using RSA key 75AD65802A0B4211
# gpg: Good signature from "Stefan Berger <[email protected]>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: B818 B9CA DF90 89C2 D5CE C66B 75AD 6580 2A0B 4211
* remotes/stefanberger/tags/pull-tpm-2018-11-15-1:
tests: tpm: Use g_test_message rather than fprintf
tpm: use loop iterator to set sts data field
Peter Maydell [Thu, 15 Nov 2018 11:46:16 +0000 (11:46 +0000)]
linux-user/sparc/signal.c: Remove dead code
Coverity complains (CID 1390847) about some dead code in
do_sigreturn(). This is an if (err) clause that can never be
true, copied from the kernel (where __get_user returns an error).
The one code path that could report an error is in the
currently commented-out pseudocode for handling FPU register
restoring, so move the if into that comment (and fix the
broken indent in the comment in the process).
(The new position for the error check is also the semantically
correct one -- we should not restore the signal mask from
the signal frame if we get an error here, so the check must
be done before set_sigmask(), not after.)
Michael Roth [Fri, 9 Nov 2018 16:13:52 +0000 (10:13 -0600)]
make-release: add skiboot .version file
This is needed to build skiboot from tarball-distributed sources
since the git data the make_release.sh script relies on to generate
it is not available.
Laurent Vivier [Tue, 13 Nov 2018 17:54:37 +0000 (18:54 +0100)]
tests/tcg/multiarch: fix 32bit linux-test on 64bit host
Fix:
TEST linux-test on i386
.../tests/tcg/multiarch/linux-test.c:201: readdir
readdir() calls getdents64() to have the list of the entries in a
directory, and getdents64() can return 64bit d_off values (with ext4,
for instance) that will not fit in the 32bit d_off field of the
readdir() dirent structure.
To avoid that, use readdir64() to use a 64bit d_off field too.
Alex Bennée [Mon, 8 Oct 2018 10:56:13 +0000 (11:56 +0100)]
.travis.yml: split MacOSX builds and reduce target list
We have reached the point where the MacOSX build was regularly timing
out. So as before I've reduced the target list to "major"
architectures to try and bring the build time down. I've added an
additional MacOSX build with the latest XCode with a minimal list of
"most likely" targets on MacOS.
Palmer Dabbelt [Fri, 9 Nov 2018 18:56:50 +0000 (10:56 -0800)]
RISC-V: Respect fences for user-only emulators
Our current fence implementation ignores fences for the user-only
configurations. This is incorrect but unlikely to manifest: it requires
multi-threaded user-only code that takes advantage of the weakness in
the host's memory model and can be inlined by TCG.
This patch simply treats fences the same way for all our emulators.
I've given it to testing as I don't want to construct a test that would
actually trigger the failure.
Our fence implementation has an additional deficiency where we map all
RISC-V fences to full fences. Now that we have a formal memory model
for RISC-V we can start to take advantage of the strength bits on our
fence instructions. This requires a bit more though, so I'm going to
split it out because the implementation is still correct without taking
advantage of these weaker fences.
Thanks to Richard Henderson for pointing out both of the issues.
s390x/pci: properly fail if the zPCI device cannot be created
Right now, errors during realize()/pre_plug/plug of the zPCI device
would result in QEMU crashing instead of failing nicely when creating
a zPCI device for a PCI device.
* remotes/vivier2/tags/linux-user-for-3.1-pull-request:
linux-user: Add support for SO_REUSEPORT
linux-user: Clean up nios2 main loop signal handling
linux-user: Don't call gdb_handlesig() before queue_signal()
linux-user: Remove dead error-checking code
Peter Maydell [Tue, 13 Nov 2018 10:52:32 +0000 (10:52 +0000)]
Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20181113' into staging
target/arm queue:
* Remove no-longer-needed workaround for small SAU regions for v8M
* Remove antique TODO comment
* MAINTAINERS: Add an entry for the 'collie' machine
* hw/arm/sysbus-fdt: Only call match_fn callback if the type matches
* Fix infinite recursion in tlbi_aa64_vmalle1_write()
* ARM KVM: fix various bugs in handling of guest debugging
* Correctly implement handling of HCR_EL2.{VI, VF}
* Hyp mode R14 is shared with User and System
* Give Cortex-A15 and -A7 the EL2 feature
* remotes/pmaydell/tags/pull-target-arm-20181113:
target/arm/cpu: Give Cortex-A15 and -A7 the EL2 feature
target/arm: Hyp mode R14 is shared with User and System
target/arm: Correctly implement handling of HCR_EL2.{VI, VF}
target/arm: Track the state of our irq lines from the GIC explicitly
Revert "target/arm: Implement HCR.VI and VF"
arm: fix aa64_generate_debug_exceptions to work with EL2
arm: use symbolic MDCR_TDE in arm_debug_target_el
tests/guest-debug: fix scoping of failcount
target/arm64: kvm debug set target_el when passing exception to guest
target/arm64: hold BQL when calling do_interrupt()
target/arm64: properly handle DBGVR RESS bits
target/arm: Fix typo in tlbi_aa64_vmalle1_write
hw/arm/sysbus-fdt: Only call match_fn callback if the type matches
MAINTAINERS: Add an entry for the 'collie' machine
target/arm: Remove antique TODO comment
target/arm: Remove workaround for small SAU regions
Peter Maydell [Tue, 13 Nov 2018 10:47:59 +0000 (10:47 +0000)]
target/arm: Hyp mode R14 is shared with User and System
Hyp mode is an exception to the general rule that each AArch32
mode has its own r13, r14 and SPSR -- it has a banked r13 and
SPSR but shares its r14 with User and System mode. We were
incorrectly implementing it as banked, which meant that on
entry to Hyp mode r14 was 0 rather than the USR/SYS r14.
We provide a new function r14_bank_number() which is like
the existing bank_number() but provides the index into
env->banked_r14[]; bank_number() provides the index to use
for env->banked_r13[] and env->banked_cpsr[].
All the points in the code that were using bank_number()
to index into env->banked_r14[] are updated for consintency:
* switch_mode() -- this is the only place where we fix
an actual bug
* aarch64_sync_32_to_64() and aarch64_sync_64_to_32():
no behavioural change as we already special-cased Hyp R14
* kvm32.c: no behavioural change since the guest can't ever
be in Hyp mode, but conceptually the right thing to do
* msr_banked()/mrs_banked(): we can never get to the case
that accesses banked_r14[] with tgtmode == ARM_CPU_MODE_HYP,
so no behavioural change
Peter Maydell [Tue, 13 Nov 2018 10:47:59 +0000 (10:47 +0000)]
target/arm: Correctly implement handling of HCR_EL2.{VI, VF}
In commit 8a0fc3a29fc2315325400 we tried to implement HCR_EL2.{VI,VF},
but we got it wrong and had to revert it.
In that commit we implemented them as simply tracking whether there
is a pending virtual IRQ or virtual FIQ. This is not correct -- these
bits cause a software-generated VIRQ/VFIQ, which is distinct from
whether there is a hardware-generated VIRQ/VFIQ caused by the
external interrupt controller. So we need to track separately
the HCR_EL2 bit state and the external virq/vfiq line state, and
OR the two together to get the actual pending VIRQ/VFIQ state.
Peter Maydell [Tue, 13 Nov 2018 10:47:59 +0000 (10:47 +0000)]
target/arm: Track the state of our irq lines from the GIC explicitly
Currently we track the state of the four irq lines from the GIC
only via the cs->interrupt_request or KVM irq state. That means
that we assume that an interrupt is asserted if and only if the
external line is set. This assumption is incorrect for VIRQ
and VFIQ, because the HCR_EL2.{VI,VF} bits allow assertion
of VIRQ and VFIQ separately from the state of the external line.
To handle this, start tracking the state of the external lines
explicitly in a CPU state struct field, as is common practice
for devices.
The complicated part of this is dealing with inbound migration
from an older QEMU which didn't have this state. We assume in
that case that the older QEMU did not implement the HCR_EL2.{VI,VF}
bits as generating interrupts, and so the line state matches
the current state in cs->interrupt_request. (This is not quite
true between commit 8a0fc3a29fc2315325400c7 and its revert, but
that commit is broken and never made it into any released QEMU
version.)
The implementation of HCR.VI and VF in that commit is not
correct -- they do not track the overall "is there a pending
VIRQ or VFIQ" status, but whether there is a pending interrupt
due to "this mechanism", ie the hypervisor having set the VI/VF
bits. The overall pending state for VIRQ and VFIQ is effectively
the logical OR of the inbound lines from the GIC with the
VI and VF bits. Commit 8a0fc3a29fc231 would result in pending
VIRQ/VFIQ possibly being lost when the hypervisor wrote to HCR.
As a preliminary to implementing the HCR.VI/VF feature properly,
revert the broken one entirely.
Alex Bennée [Tue, 13 Nov 2018 10:47:59 +0000 (10:47 +0000)]
arm: fix aa64_generate_debug_exceptions to work with EL2
The test was incomplete and incorrectly caused debug exceptions to be
generated when returning to EL2 after a failed attempt to single-step
an EL1 instruction. Fix this while cleaning up the function a little.
projects / qemu.git / log