Peter Maydell [Mon, 7 Jan 2019 15:23:47 +0000 (15:23 +0000)]
cpus.c: Fix race condition in cpu_stop_current()
We use cpu_stop_current() to ensure the current CPU has stopped
from places like qemu_system_reset_request(). Unfortunately its
current implementation has a race. It calls qemu_cpu_stop(),
which sets cpu->stopped to true even though the CPU hasn't
actually stopped yet. The main thread will look at the flags
set by qemu_system_reset_request() and call pause_all_vcpus().
pause_all_vcpus() waits for every cpu to have cpu->stopped true,
so it can continue (and we will start the system reset operation)
before the vcpu thread has got back to its top level loop.
Instead, just set cpu->stop and call cpu_exit(). This will
cause the vcpu to exit back to the top level loop, and there
(as part of the wait_io_event code) it will call qemu_cpu_stop().
This fixes bugs where the reset request appeared to be ignored
or the CPU misbehaved because the reset operation started
to change vcpu state while the vcpu thread was still using it.
Thomas Huth [Mon, 7 Jan 2019 15:23:47 +0000 (15:23 +0000)]
MAINTAINERS: Add ARM-related files for hw/[misc|input|timer]/
Some of the files in hw/input/, hw/misc/ and hw/timer/ are only
used by one of the ARM machines, so we can assign these files to
the corresponding boards.
The generic loader device (-device loader,file=kernel.bin) can be used
to load a kernel instead of the -kernel option. Some boards have flash
memory (pflash) that is set via the -pflash or -drive options.
Allow starting QEMU without the -kernel option to accommodate these
scenarios.
Luc Michel [Mon, 7 Jan 2019 15:23:46 +0000 (15:23 +0000)]
gdbstub: add multiprocess extension support
Add multiprocess extension support by enabling multiprocess mode when
the peer requests it, and by replying that we actually support it in the
qSupported reply packet.
Luc Michel [Mon, 7 Jan 2019 15:23:46 +0000 (15:23 +0000)]
gdbstub: gdb_set_stop_cpu: ignore request when process is not attached
When gdb_set_stop_cpu() is called with a CPU associated to a process
currently not attached by the GDB client, return without modifying the
stop CPU. Otherwise, GDB gets confused if it receives packets with a
thread-id it does not know about.
Luc Michel [Mon, 7 Jan 2019 15:23:46 +0000 (15:23 +0000)]
gdbstub: processes initialization on new peer connection
When a new connection is established, we set the first process to be
attached, and the others detached. The first CPU of the first process
is selected as the current CPU.
Luc Michel [Mon, 7 Jan 2019 15:23:46 +0000 (15:23 +0000)]
gdbstub: add multiprocess support to Xfer:features:read:
Change the Xfer:features:read: packet handling to support the
multiprocess extension. This packet is used to request the XML
description of the CPU. In multiprocess mode, different descriptions can
be sent for different processes.
This function now takes the process to send the description for as a
parameter, and use a buffer in the process structure to store the
generated description.
It takes the first CPU of the process to generate the description.
Luc Michel [Mon, 7 Jan 2019 15:23:46 +0000 (15:23 +0000)]
gdbstub: add multiprocess support to 'sC' packets
Change the sC packet handling to support the multiprocess extension.
Instead of returning the first thread, we return the first thread of the
current process.
Luc Michel [Mon, 7 Jan 2019 15:23:45 +0000 (15:23 +0000)]
gdbstub: add multiprocess support to 'H' and 'T' packets
Add a couple of helper functions to cope with GDB threads and processes.
The gdb_get_process() function looks for a process given a pid.
The gdb_get_cpu() function returns the CPU corresponding to the (pid,
tid) pair given as parameters.
The read_thread_id() function parses the thread-id sent by the peer.
This function supports the multiprocess extension thread-id syntax. The
return value specifies if the parsing failed, or if a special case was
encountered (all processes or all threads).
Use them in 'H' and 'T' packets handling to support the multiprocess
extension.
Luc Michel [Mon, 7 Jan 2019 15:23:45 +0000 (15:23 +0000)]
gdbstub: add multiprocess support to '?' packets
The gdb_get_cpu_pid() function does the PID lookup for the given CPU. It
checks if the CPU is a direct child of a CPU cluster. If it is, the
returned PID is the cluster ID plus one (cluster IDs start at 0, GDB
PIDs at 1). When the CPU is not a child of such a container, the PID of
the default process is returned.
The gdb_fmt_thread_id() function generates the string to be used to identify
a given thread, in a response packet for the peer. This function
supports generating thread IDs when multiprocess mode is enabled (in the
form `p<pid>.<tid>').
Luc Michel [Mon, 7 Jan 2019 15:23:45 +0000 (15:23 +0000)]
gdbstub: introduce GDB processes
Add a structure GDBProcess that represents processes from the GDB
semantic point of view.
CPUs can be split into different processes, by grouping them under
different cpu-cluster objects. Each occurrence of a cpu-cluster object
implies the existence of the corresponding process in the GDB stub. The
GDB process ID is derived from the corresponding cluster ID as follows:
GDB PID = cluster ID + 1
This is because PIDs -1 and 0 are reserved in GDB and cannot be used by
processes.
A default process is created to handle CPUs that are not in a cluster.
This process gets the PID of the last process PID + 1.
target/arm: SVE brk[ab] merging does not have s bit
While brk[ab] zeroing has a flags setting option, the merging variant
does not. Retain the same argument structure, to share expansion but
force the flag zero and do not decode bit 22.
Use "register" TBFLAG_ANY to indicate shared state between
A32 and A64, and "registers" TBFLAG_A32 & TBFLAG_A64 for
fields that are specific to the given cpu state.
Move ARM_TBFLAG_BE_DATA to shared state, instead of its current
placement within "Bit usage when in AArch32 state".
Peter Maydell [Mon, 7 Jan 2019 11:55:52 +0000 (11:55 +0000)]
Merge remote-tracking branch 'remotes/ericb/tags/pull-nbd-2019-01-05' into staging
nbd patches for 2019-01-05
Error and trace improvements in NBD code, such as less noise for
common disconnect scenarios.
- Vladimir Sementsov-Ogievskiy: 0/3 nbd-client: drop extra error noise
- Eric Blake: portions of 0/22 nbd: add qemu-nbd --list
# gpg: Signature made Sat 05 Jan 2019 13:58:54 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-2019-01-05:
nbd/client: Drop pointless buf variable
qemu-nbd: Fail earlier for -c/-d on non-linux
nbd/client: More consistent error messages
nbd: Document timeline of various features
qemu-nbd: Use program name in error messages
block/nbd-client: use traces instead of noisy error_report_err
nbd/client: Trace all server option error messages
nbd: publish _lookup functions
Eric Blake [Sat, 15 Dec 2018 13:53:10 +0000 (07:53 -0600)]
nbd/client: Drop pointless buf variable
There's no need to read into a temporary buffer (oversized
since commit 7d3123e1) followed by a byteswap into a uint64_t
to check for a magic number via memcmp(), when the code
immediately below demonstrates reading into the uint64_t then
byteswapping in place and checking for a magic number via
integer math. What's more, having a different error message
when the server's first reply byte is 0 is unusual - it's no
different from any other wrong magic number, and we already
detected short reads. That whole strlen() issue has been
present and useless since commit 1d45f8b5 in 2010; perhaps it
was leftover debugging (since the correct magic number happens
to be ASCII)? Make the error messages more consistent and
detailed while touching things.
Eric Blake [Sat, 15 Dec 2018 13:53:08 +0000 (07:53 -0600)]
qemu-nbd: Fail earlier for -c/-d on non-linux
Connecting to a /dev/nbdN device is a Linux-specific action.
We were already masking -c and -d from 'qemu-nbd --help' on
non-linux. However, while -d fails with a sensible error
message, it took hunting through a couple of files to prove
that. What's more, the code for -c doesn't fail until after
it has created a pthread and tried to open a device - possibly
even printing an error message with %m on a non-Linux platform
in spite of the comment that %m is glibc-specific. Make the
failure happen sooner, then get rid of stubs that are no
longer needed because of the early exits.
While at it: tweak the blank newlines in --help output to be
consistent, whether or not built on Linux.
Eric Blake [Sat, 15 Dec 2018 13:53:07 +0000 (07:53 -0600)]
nbd/client: More consistent error messages
Consolidate on using decimal (not hex), on outputting the
option reply name (not just value), and a consistent comma between
clauses, when the client reports protocol discrepancies from the
server. While it won't affect normal operation, it makes
debugging additions easier.
Eric Blake [Sat, 15 Dec 2018 13:53:04 +0000 (07:53 -0600)]
nbd: Document timeline of various features
It can be useful to figure out which NBD protocol features are
exposed by a server, as well as what features a client will
take advantage of if available, for a given qemu release. It's
not always precise to base features on version numbers (thanks
to downstream backports), but any documentation is better than
making users search through git logs themselves.
This patch originally stemmed from a request to document that
pristine 3.0 has a known bug where NBD_OPT_LIST_META_CONTEXT
with 0 queries forgot to advertise an available
"qemu:dirty-bitmap" context, but documenting bugs like this (or
the fact that 3.0 also botched NBD_CMD_CACHE) gets to be too
much details, especially since buggy releases will be less
likely connection targets over time. Instead, I chose to just
remind users to check stable release branches.
block/nbd-client: use traces instead of noisy error_report_err
Reduce extra noise of nbd-client, change 083 correspondingly.
In various commits (be41c100 in 2.10, f140e300 in 2.11, 78a33ab
in 2.12), we added spots where qemu as an NBD client would report
problems communicating with the server to stderr, because there
was no where else to send the error to. However, this is racy,
particularly since the most common source of these errors is when
either the client or the server abruptly hangs up, leaving one
coroutine to report the error only if it wins (or loses) the
race in attempting the read from the server before another
thread completes its cleanup of a protocol error that caused the
disconnect in the first place. The race is also apparent in the
fact that differences in the flush behavior of the server can
alter the frequency of encountering the race in the client (see
commit 6d39db96).
Rather than polluting stderr, it's better to just trace these
situations, for use by developers debugging a flaky connection,
particularly since the real error that either triggers the abrupt
disconnection in the first place, or that results from the EIO
when a request can't receive a reply, DOES make it back to the
user in the normal Error propagation channels.
Eric Blake [Tue, 18 Dec 2018 22:57:13 +0000 (16:57 -0600)]
nbd/client: Trace all server option error messages
Not all servers send free-form text alongside option error replies, but
for servers that do (such as qemu), we pass the server's message as a
hint alongside our own error reporting. However, it would also be
useful to trace such server messages, since we can't guarantee how the
hint may be consumed.
These functions are used for formatting pretty trace points. We are
going to add some in block/nbd-client, so, let's publish all these
functions at once. Note, that nbd_reply_type_lookup is already
published, and constants, "named" by these functions live in
include/block/nbd.h too.
Li Qiang [Wed, 21 Nov 2018 05:10:25 +0000 (21:10 -0800)]
fw_cfg: Fix -boot reboot-timeout error checking
fw_cfg_reboot() gets option parameter "reboot-timeout" with
qemu_opt_get(), then converts it to an integer by hand. It neglects to
check that conversion for errors, and fails to reject negative values.
Positive values above the limit get reported and replaced by the limit.
This patch checks for conversion errors properly, and reject all values
outside 0...0xffff.
Li Qiang [Wed, 21 Nov 2018 05:10:24 +0000 (21:10 -0800)]
fw_cfg: Fix -boot bootsplash error checking
fw_cfg_bootsplash() gets option parameter "splash-time"
with qemu_opt_get(), then converts it to an integer by hand.
It neglects to check that conversion for errors. This is
needlessly complicated and error-prone. But as "splash-time
not specified" is not the same as "splash-time=T" for any T,
we need use qemu_opt_get() to check if splash time exists.
This patch also make the qemu exit when finding or loading
splash file failed.
Li Qiang [Thu, 1 Nov 2018 06:02:28 +0000 (23:02 -0700)]
fw_cfg: Improve error message when can't load splash file
read_splashfile() reports "failed to read splash file" without
further details. Get the details from g_file_get_contents(), and
include them in the error message. Also remove unnecessary 'res'
variable.
Peter Maydell [Fri, 4 Jan 2019 13:22:51 +0000 (13:22 +0000)]
Merge remote-tracking branch 'remotes/stefanha/tags/block-pull-request' into staging
Pull request
Bug fixes for the .dmg image file format.
# gpg: Signature made Fri 04 Jan 2019 11:21:18 GMT
# gpg: using RSA key 9CA4ABB381AB73C8
# gpg: Good signature from "Stefan Hajnoczi <[email protected]>"
# gpg: aka "Stefan Hajnoczi <[email protected]>"
# Primary key fingerprint: 8695 A8BF D3F9 7CDA AC35 775A 9CA4 ABB3 81AB 73C8
* remotes/stefanha/tags/block-pull-request:
dmg: don't skip zero chunk
dmg: use enumeration type instead of hard coding number
dmg: fix binary search
dmg: Fixing wrong dmg block type value for block terminator.
yuchenlin [Thu, 3 Jan 2019 11:47:00 +0000 (19:47 +0800)]
dmg: don't skip zero chunk
The dmg file has many tables which describe: "start from sector XXX to
sector XXX, the compression method is XXX and where the compressed data
resides on".
Each sector in the expanded file should be covered by a table. The table
will describe the offset of compressed data (or raw depends on the type)
in the dmg.
For example:
[-----------The expanded file------------]
[---bzip table ---]/* zeros */[---zlib---]
^
| if we want to read this sector.
we will find bzip table which contains this sector, and get the
compressed data offset, read it from dmg, uncompress it, finally write to
expanded file.
If we skip zero chunk (table), some sector cannot find the table which
will cause search_chunk() return s->n_chunks, dmg_read_chunk() return -1
and finally causing dmg_co_preadv() return EIO.
See:
[-----------The expanded file------------]
[---bzip table ---]/* zeros */[---zlib---]
^
| if we want to read this sector.
Oops, we cannot find the table contains it...
In the original implementation, we don't have zero table. When we try to
read sector inside the zero chunk. We will get EIO, and skip reading.
After this patch, we treat zero chunk the same as ignore chunk, it will
directly write zero and avoid some sector may not find the table.
Julio Faracco [Fri, 28 Dec 2018 14:50:55 +0000 (12:50 -0200)]
dmg: Fixing wrong dmg block type value for block terminator.
This is a trivial patch to fix a wrong value for block terminator.
The old value was 0x7fffffff which is wrong. It was not affecting the
code because QEMU dmg block is not handling block terminator right now.
Neverthless, it should be fixed.
Peter Maydell [Fri, 4 Jan 2019 10:11:18 +0000 (10:11 +0000)]
Merge remote-tracking branch 'remotes/amarkovic/tags/mips-queue-december-2018-v3' into staging
MIPS queue for December 2018 - v3
# gpg: Signature made Thu 03 Jan 2019 16:53:47 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-december-2018-v3: (44 commits)
tests/tcg: mips: Test R5900 three-operand MADDU1
tests/tcg: mips: Test R5900 three-operand MADDU
tests/tcg: mips: Test R5900 three-operand MADD1
tests/tcg: mips: Test R5900 three-operand MADD
disas: nanoMIPS: Add a note on documentation
disas: nanoMIPS: Reorder declarations and definitions of gpr decoders
disas: nanoMIPS: Comment the decoder of 'gpr1' gpr encoding type
disas: nanoMIPS: Rename the decoder of 'gpr1' gpr encoding type
disas: nanoMIPS: Comment the decoder of 'gpr2.reg2' gpr encoding type
disas: nanoMIPS: Rename the decoder of 'gpr2.reg2' gpr encoding type
disas: nanoMIPS: Comment the decoder of 'gpr2.reg1' gpr encoding type
disas: nanoMIPS: Rename the decoder of 'gpr2.reg1' gpr encoding type
disas: nanoMIPS: Comment the decoder of 'gpr4.zero' gpr encoding type
disas: nanoMIPS: Rename the decoder of 'gpr4.zero' gpr encoding type
disas: nanoMIPS: Comment the decoder of 'gpr4' gpr encoding type
disas: nanoMIPS: Rename the decoder of 'gpr4' gpr encoding type
disas: nanoMIPS: Comment the decoder of 'gpr3.src.store' gpr encoding type
disas: nanoMIPS: Rename the decoder of 'gpr3.src.store' gpr encoding type
disas: nanoMIPS: Comment the decoder of 'gpr3' gpr encoding type
disas: nanoMIPS: Rename the decoder of 'gpr3' gpr encoding type
...
Fix order of extraction function invocations so that extraction
goes from MSB side to LSB side of the given instruction coding
content. This is desireable because of consistency and easier
visual spotting of errors.
After this patch, all such invocations should be in the desired
order.
Fix order of extraction function invocations so that extraction
goes from MSB side to LSB side of the given instruction coding
content. This is desireable because of consistency and easier
visual spotting of errors.
target/mips: MXU: Add handlers for logic instructions
Add translation handlers for four logic MXU instructions.
It should be noted that there is an error in MXU documentation (dated
June 2017) regarding opcodes for this group of instructions. This was
confirmed by running tests on hardware, and also by looking up other
related public source trees (binutils, Android NDK). In initial MXU
patches to QEMU, opcodes for MXU logic instructions were created to
be in accordance with the MXU documentation, therefore the error from
was propagated. This patch corrects that, changing the involved code.
Besides that, as MXU was designed and implemented only for 32-bit
CPUs, corresponding preprosessor conditions were added around MXU
code, which allows more flexible implementation of MXU handlers.
target/mips: MXU: Add generic naming for optn2 constants
Add generic naming involving generig suffixes OPTN0, OPTN1, OPTN2,
OPTN3 for four optn2 constants. Existing suffixes WW, LW, HW, XW
are not quite appropriate for some instructions using optn2.
target/mips: MXU: Add missing opcodes/decoding for LX* instructions
Add missing opcodes and decoding engine for LXB, LXH, LXW, LXBU,
and LXHU instructions. They were for some reason forgotten in
previous commits. The MXU opcode list and decoding engine should
be now complete.
Paul Burton [Thu, 27 Dec 2018 15:32:11 +0000 (16:32 +0100)]
atomics: Set ATOMIC_REG_SIZE=8 for MIPS n32
ATOMIC_REG_SIZE is currently defined as the default sizeof(void *) for
all MIPS host builds, including those using the n32 ABI. n32 is the
MIPS64 ILP32 ABI and as such tcg/mips/tcg-target.h defines
TCG_TARGET_REG_BITS as 64 for n32 builds. If we attempt to build QEMU
for an n32 host with support for a 64b target architecture then
TCG_OVERSIZED_GUEST is 0 and accel/tcg/cputlb.c attempts to use
atomic_* functions. This fails because ATOMIC_REG_SIZE is 4, causing
the calls to QEMU_BUILD_BUG_ON(sizeof(*ptr) > ATOMIC_REG_SIZE) in the
various atomic_* functions to generate errors.
Fix this by defining ATOMIC_REG_SIZE as 8 for all MIPS64 builds, which
will cover both n32 (ILP32) & n64 (LP64) ABIs in much the same was as
we already do for x86_64/x32.
MAINTAINERS: Add Aleksandar Rikalo as a reviewer for MIPS content
Add Aleksandar Rikalo as a reviewer for MIPS content. Aleksandar
brings to us more than six years of experience in working on a variety
of development tools for MIPS architectures, and will greatly help
QEMU community understand and support intricacies of MIPS better.
Future nanoMIPS user mode will also have its .mak file, and
because of that "*mips*" was used instead of "mips*" as a
shorthand in the new item in MAINTAINERS.
Peter Maydell [Thu, 3 Jan 2019 13:26:30 +0000 (13:26 +0000)]
Merge remote-tracking branch 'remotes/palmer/tags/riscv-for-master-3.2-part1' into staging
RISC-V Changes for 3.2, Part 1
This pull request contains the first set of RISC-V patches I'd like to
target for the 3.2 development cycle. It's really just a collection of
bug fixes with one major new feature: PCIe can now be attached to RISC-V
guests.
This has passed my usual test of booting the latest Linux RC into a
Fedora disk image on the virt machine.
# gpg: Signature made Fri 21 Dec 2018 16:01:29 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/palmer/tags/riscv-for-master-3.2-part1:
MAINTAINERS: Mark RISC-V as Supported
riscv/cpu: use device_class_set_parent_realize
target/riscv/pmp.c: Fix pmp_decode_napot()
sifive_uart: Implement interrupt pending register
RISC-V: Enable second UART on sifive_e and sifive_u
RISC-V: Fix PLIC pending bitfield reads
RISC-V: Fix CLINT timecmp low 32-bit writes
RISC-V: Add hartid and \n to interrupt logging
sifive_u: Set 'clock-frequency' DT property for SiFive UART
sifive_u: Add clock DT node for GEM ethernet
riscv: Enable VGA and PCIE_VGA
hw/riscv/virt: Connect the gpex PCIe
hw/riscv/virt: Adjust memory layout spacing
hw/riscv/virt: Increase the number of interrupts
Free the argument register only after we have verified that the
temporary is not already in that register. This case is likely
now that we are back propagating the preferred register.
Use this to notice the opcodes that exit the TB, which implies
that local temps are really dead and need not be synced.
Previously we so marked the true end of the TB, but that was
immediately overwritten by the la_bb_end invoked by any
TCG_OPF_BB_END opcode, like exit_tb.
No need for a "tcg_" prefix for a static function; we already
have another "la_" prefix for indicating liveness analysis.
Pass in nb_globals and nb_temps, as we will already have them
in registers for other loops within the parent function.
tcg: Improve register allocation for matching constraints
Try harder to honor the output_pref. When we're forced to allocate
a second register for the input, it does not need to use the input
constraint; that will be honored by the register we allocate for the
output and a move is already required.
Allocate storage for, but do not yet fill in, per-opcode
preferences for the output operands. Pass it in to the
register allocation routines for output operands.
projects / qemu.git / log