Atish Patra [Thu, 7 Jan 2021 09:11:27 +0000 (01:11 -0800)]
RISC-V: Place DTB at 3GB boundary instead of 4GB
Currently, we place the DTB at 2MB from 4GB or end of DRAM which ever is
lesser. However, Linux kernel can address only 1GB of memory for RV32.
Thus, it can not map anything beyond 3GB (assuming 2GB is the starting address).
As a result, it can not process DT and panic if opensbi dynamic firmware
is used. While at it, place the DTB further away to avoid in memory placement
issues in future.
Fix this by placing the DTB at 16MB from 3GB or end of DRAM whichever is lower.
Target description is not currently implemented in RISC-V
architecture. Thus GDB won't set it properly when attached.
The patch implements the target description response.
Xuzhou Cheng [Wed, 23 Dec 2020 02:00:25 +0000 (10:00 +0800)]
hw/block: m25p80: Implement AAI-WP command support for SST flashes
Auto Address Increment (AAI) Word-Program is a special command of
SST flashes. AAI-WP allows multiple bytes of data to be programmed
without re-issuing the next sequential address location.
* remotes/kraxel/tags/audio-20210115-pull-request: (30 commits)
audio: space prohibited between function name and parenthesis'('
audio: Suspect code indent for conditional statements
audio: Don't use '%#' in format strings
audio: Fix lines over 90 characters
audio: foo* bar" should be "foo *bar".
audio: Add spaces around operator/delete redundant spaces
audio: Add braces for statements/fix braces' position
dsoundaudio: fix log message
dsoundaudio: enable f32 audio sample format
dsoundaudio: rename dsound_open()
dsoundaudio: replace GetForegroundWindow()
paaudio: send recorded data in smaller chunks
paaudio: limit minreq to 75% of audio timer_rate
paaudio: comment bugs in functions qpa_init_*
paaudio: remove unneeded code
paaudio: wait until the playback stream is ready
paaudio: wait for PA_STREAM_READY in qpa_write()
paaudio: avoid to clip samples multiple times
audio: remove remaining unused plive code
sdlaudio: enable (in|out).mixing-engine=off
...
* remotes/kraxel/tags/ui-20210115-pull-request:
vnc: add support for extended desktop resize
vnc: move initialization to framebuffer_update_request
vnc: move check into vnc_cursor_define
vnc: Fix a memleak in vnc_display_connect()
ui: add support for remote power control to VNC server
vnc: fix unfinalized tlscreds for VncDisplay
ui/gtk: update monitor interval on egl displays
ui/gtk: expose gd_monitor_update_interval
ui/gtk: limit virtual console max update interval
ui/gtk: rename variable window to widget
ui/gtk: don't try to redefine SI prefixes
Peter Maydell [Fri, 15 Jan 2021 13:22:54 +0000 (13:22 +0000)]
Merge remote-tracking branch 'remotes/philmd-gitlab/tags/mips-20210114' into staging
MIPS patches queue
- Simplify CPU/ISA definitions
- Various maintenance code movements in translate.c
- Convert part of the MSA ASE instructions to decodetree
- Convert some instructions removed from Release 6 to decodetree
- Remove deprecated 'fulong2e' machine alias
# gpg: Signature made Thu 14 Jan 2021 16:16:29 GMT
# gpg: using RSA key FAABE75E12917221DCFD6BB2E3E32C2CDEADC0DE
# gpg: Good signature from "Philippe Mathieu-Daudé (F4BUG) <[email protected]>" [full]
# Primary key fingerprint: FAAB E75E 1291 7221 DCFD 6BB2 E3E3 2C2C DEAD C0DE
* remotes/philmd-gitlab/tags/mips-20210114: (69 commits)
docs/system: Remove deprecated 'fulong2e' machine alias
target/mips: Remove vendor specific CPU definitions
target/mips: Remove CPU_NANOMIPS32 definition
target/mips: Remove CPU_R5900 definition
target/mips: Convert Rel6 LL/SC opcodes to decodetree
target/mips: Convert Rel6 LLD/SCD opcodes to decodetree
target/mips: Convert Rel6 LDL/LDR/SDL/SDR opcodes to decodetree
target/mips: Convert Rel6 LWLE/LWRE/SWLE/SWRE opcodes to decodetree
target/mips: Convert Rel6 LWL/LWR/SWL/SWR opcodes to decodetree
target/mips: Convert Rel6 CACHE/PREF opcodes to decodetree
target/mips: Convert Rel6 COP1X opcode to decodetree
target/mips: Convert Rel6 Special2 opcode to decodetree
target/mips: Remove now unreachable LSA/DLSA opcodes code
target/mips: Introduce decodetree helpers for Release6 LSA/DLSA opcodes
target/mips: Introduce decodetree helpers for MSA LSA/DLSA opcodes
target/mips: Extract LSA/DLSA translation generators
target/mips: Use decode_ase_msa() generated from decodetree
target/mips: Introduce decode tree bindings for MSA ASE
target/mips: Pass TCGCond argument to MSA gen_check_zero_element()
target/mips: Extract MSA translation routines
...
Volker Rümelin [Sun, 10 Jan 2021 10:02:36 +0000 (11:02 +0100)]
dsoundaudio: replace GetForegroundWindow()
GetForegroundWindow() doesn't necessarily return the own window
handle. It just returns a handle to the currently active window
and can even return NULL. At the time dsound_open() gets called
the active window is most likely the shell window and not the
QEMU window.
Replace GetForegroundWindow() with GetDesktopWindow() which
always returns a valid window handle, and at the same time
replace the DirectSound buffer flag DSBCAPS_STICKYFOCUS with
DSBCAPS_GLOBALFOCUS where Windows only expects a valid window
handle for DirectSound function SetCooperativeLevel(). The
Microsoft online docs for IDirectSound::SetCooperativeLevel
recommend this in the remarks.
This fixes a bug where you can't hear sound from the guest.
To reproduce start qemu with -machine pcspk-audiodev=audio0
-device intel-hda -device hda-duplex,audiodev=audio0
-audiodev dsound,id=audio0,out.mixing-engine=off
from a shell and start audio playback with the hda device in the
guest. The guest will be silent. To hear guest audio you have to
activate the shell window once.
Volker Rümelin [Sun, 10 Jan 2021 10:02:35 +0000 (11:02 +0100)]
paaudio: send recorded data in smaller chunks
Tell PulseAudio to send recorded audio data in smaller chunks
than timer_period, so there's a good chance that qemu can read
recorded audio data every time it looks for new data.
PulseAudio tries to send buffer updates at a fragsize / 2 rate.
With fragsize = timer_period / 2 * 3 the update rate is 75% of
timer_period. The lower limit for the recording buffer size
maxlength is fragsize * 2.
Volker Rümelin [Sun, 10 Jan 2021 10:02:34 +0000 (11:02 +0100)]
paaudio: limit minreq to 75% of audio timer_rate
Currently with the playback buffer attribute minreq = -1 and flag
PA_STREAM_EARLY_REQUESTS PulseAudio uses minreq = tlength / 4.
To improve audio playback with larger PulseAudio server side
buffers, limit minreq to a maximum of 75% of audio timer_rate.
That way there is a good chance qemu receives a stream buffer
size update before it tries to write data to the playback stream.
Volker Rümelin [Sun, 10 Jan 2021 10:02:33 +0000 (11:02 +0100)]
paaudio: comment bugs in functions qpa_init_*
The audio buffer size in audio/paaudio.c is typically larger
than expected. Just comment the bugs in qpa_init_in() and
qpa_init_out() for now. Fixing these bugs may break glitch free
audio playback with fine tuned user audio settings.
Volker Rümelin [Sun, 10 Jan 2021 10:02:32 +0000 (11:02 +0100)]
paaudio: remove unneeded code
Commit baea032ec7 "audio/paaudio: fix ignored buffer_length setting"
added code to handle buffer_length defaults. This was unnecessary
because the audio_buffer_* functions in audio/audio.c already handle
this. Remove the unneeded code.
Volker Rümelin [Sun, 10 Jan 2021 10:02:31 +0000 (11:02 +0100)]
paaudio: wait until the playback stream is ready
Don't call pa_stream_writable_size() in qpa_get_buffer_out()
before the playback stream is ready. This prevents a lot of the
following pulseaudio error messages.
pulseaudio: pa_stream_writable_size failed
pulseaudio: Reason: Bad state
To reproduce start qemu with
-parallel none -device gus,audiodev=audio0 -audiodev pa,id=audio0
Volker Rümelin [Sun, 10 Jan 2021 10:02:30 +0000 (11:02 +0100)]
paaudio: wait for PA_STREAM_READY in qpa_write()
Don't call pa_stream_writable_size() in qpa_write() before the
playback stream is ready. This prevents a lot of the following
pulseaudio error messages.
pulseaudio: pa_stream_writable_size failed
pulseaudio: Reason: Bad state
To reproduce start qemu with
-parallel none -device gus,audiodev=audio0
-audiodev pa,id=audio0,out.mixing-engine=off
Volker Rümelin [Sun, 10 Jan 2021 10:02:29 +0000 (11:02 +0100)]
paaudio: avoid to clip samples multiple times
The pulseaudio backend currently converts, clips and copies audio
playback samples in the mixing-engine sample buffer multiple
times.
In qpa_get_buffer_out() the function pa_stream_begin_write()
returns a rather large buffer and this allows audio_pcm_hw_run_out()
in audio/audio.c to copy all samples in the mixing-engine buffer
to the pulse audio buffer. Immediately after copying, qpa_write()
notices with a call to pa_stream_writable_size() that pulse audio
only needs a smaller part of the copied samples and ignores the
rest. This copy and ignore process happens several times for each
audio sample.
To fix this behaviour, call pa_stream_writable_size() in
qpa_get_buffer_out() to limit the number of samples
audio_pcm_hw_run_out() will convert. With this change the
pulseaudio pcm_ops functions put_buffer_out and write are no
longer identical and a separate qpa_put_buffer_out is needed.
Volker Rümelin [Sun, 10 Jan 2021 10:02:25 +0000 (11:02 +0100)]
sdlaudio: add recording functions
Add audio recording functions. SDL 2.0.5 or later is required to
use the recording functions. Playback continues to work with
earlier SDL 2.0 versions.
Volker Rümelin [Sun, 10 Jan 2021 10:02:21 +0000 (11:02 +0100)]
sdlaudio: always clear the sample buffer
Always fill the remaining audio callback buffer with silence.
SDL 2.0 doesn't initialize the audio callback buffer. This was
an incompatible change compared to SDL 1.2. For reference read
the SDL 1.2 to 2.0 migration guide.
Volker Rümelin [Sun, 10 Jan 2021 10:02:20 +0000 (11:02 +0100)]
sdlaudio: don't start playback in init routine
Every emulated audio device has a way to enable audio playback. Don't
start playback until the guest enables the audio device. This patch
keeps the SDL2 device pause state in sync with hw->enabled.
Currently there is a crackling noise with SDL2 audio playback.
Commit bcf19777df: "audio/sdlaudio: Allow audio playback with
SDL2" already mentioned the crackling noise.
Add an out.buffer-count option to give users a chance to select
sane settings for glitch free audio playback. The idea was taken
from the coreaudio backend.
The in.buffer-count option will be used with one of the next
patches.
Gerd Hoffmann [Tue, 12 Jan 2021 13:41:20 +0000 (14:41 +0100)]
vnc: add support for extended desktop resize
The extended desktop resize encoding adds support for (a) clients
sending resize requests to the server, and (b) multihead support.
This patch implements (a). All resize requests are rejected by qemu.
Qemu can't resize the framebuffer on its own, this is in the hands of
the guest, so all qemu can do is forward the request to the guest.
Should the guest actually resize the framebuffer we can notify the vnc
client later with a separate message.
This requires support in the display device. Works with virtio-gpu.
Gerd Hoffmann [Tue, 12 Jan 2021 13:41:19 +0000 (14:41 +0100)]
vnc: move initialization to framebuffer_update_request
qemu sends various state info like current cursor shape to newly connected
clients in response to a set_encoding message. This is not correct according
to the rfb spec. Send that information in response to a full (incremental=0)
framebuffer update request instead. Also send the resize information
unconditionally, not only in case of an actual server-side change.
This makes the qemu vnc server conform to the spec and allows clients to
request the complete vnc server state without reconnect.
ui: add support for remote power control to VNC server
The "XVP" (Xen VNC Proxy) extension defines a mechanism for a VNC client
to issue power control requests to trigger graceful shutdown, reboot, or
hard reset.
This option is not enabled by default, since we cannot assume that users
with VNC access implicitly have administrator access to the guest OS.
Thus is it enabled with a boolean "power-control" option e.g.
-vnc :1,power-control=on
While, QEMU can easily support shutdown and reset, there's no easy way
to wire up reboot support at this time. In theory it could be done by
issuing a shutdown, followed by a reset, but there's no convenient
wiring for such a pairing in QEMU. It also isn't possible to have the
VNC server directly talk to QEMU guest agent, since the agent chardev is
typically owned by an external mgmt app.
Signed-off-by: Daniel P. Berrangé <[email protected]>
[ kraxel: rebase to master ]
[ kraxel: add missing break ]
Zihao Chang [Mon, 11 Jan 2021 13:19:11 +0000 (21:19 +0800)]
vnc: fix unfinalized tlscreds for VncDisplay
In vnc_display_open(), if tls-creds is enabled, do object_ref(object
ref 1->2) for tls-creds. While in vnc_display_close(), object_unparent
sets object ref to 1(2->1) and unparent the object for root.
Problem:
1. the object can not be found from the objects_root, while the object
is not finalized.
2. the qemu_opts of tls-creds(id: creds0) is not deleted, so new tls
object with the same id(creds0) can not be delete & add.
Nikola Pavlica [Thu, 14 Jan 2021 14:01:53 +0000 (15:01 +0100)]
ui/gtk: update monitor interval on egl displays
When running QEMU's GTK UI without EGL or OGL, the
gd_monitor_update_interval function gets executed and the display refresh
rate gets updated accordingly. However, when using EGL or just regular
OGL, the function never gets executed.
Which is why I decided that the function should be in gd_egl_refresh
where the display output gets updated, in the same vain as how it's done
for normal GTK UIs (aka. those without EGL) - in it's display refresh
function.
Since the gd_monitor_update_interval function now is exposed, we are
going to use it to update the refresh rate.
Nikola Pavlica [Thu, 14 Jan 2021 14:01:52 +0000 (15:01 +0100)]
ui/gtk: expose gd_monitor_update_interval
The gd_egl_refresh function, as the name suggests, is responsible for
refreshing displays when using EGL graphics with QEMU's GTK UI. This is
a perfect candidate for a function to update the refresh rate in.
Since gd_monitor_update_interval is inaccessible from the gd_egl_refresh
function, we need to expose/globalize it in the include/ui/gtk.h file.
Volker Rümelin [Sun, 13 Dec 2020 16:57:24 +0000 (17:57 +0100)]
ui/gtk: limit virtual console max update interval
Limit the virtual console maximum update interval to
GUI_REFRESH_INTERVAL_DEFAULT. This papers over a integer
overflow bug in gtk3 on Windows where the reported monitor
refresh frequency can be much smaller than the real refresh
frequency.
The gtk bug report can be found here:
https://gitlab.gnome.org/GNOME/gtk/-/issues/3394
On my Windows 10 system gtk reports a monitor refresh rate of
1.511Hz instead of 60.031Hz and slows down the screen update
rate in qemu to a crawl. Provided you are affected by the gtk
bug on Windows, these are the steps to reproduce the issue:
Start qemu with -display gtk and activate all qemu virtual
consoles and notice the reduced qemu refresh rate. Activating
all virtual consoles is necessary, because gui_update() in
ui/console.c uses the minimum of all display change listeners
update interval and not yet activated virtual consoles report
the default update interval (30ms).
Volker Rümelin [Sun, 13 Dec 2020 16:57:22 +0000 (17:57 +0100)]
ui/gtk: don't try to redefine SI prefixes
Redefining SI prefixes is always wrong. 1s has per definition
1000ms. Remove the misnamed named constant and replace it with
a comment explaining the frequency to period conversion in two
simple steps. Now you can cancel out the unit mHz in the comment
with the implicit unit mHz in refresh_rate_millihz and see why
the implicit unit ms for update_interval remains.
Depending on the client activity, the server can be asked to open a huge
number of file descriptors and eventually hit RLIMIT_NOFILE. This is
currently mitigated using a reclaim logic : the server closes the file
descriptors of idle fids, based on the assumption that it will be able
to re-open them later. This assumption doesn't hold of course if the
client requests the file to be unlinked. In this case, we loop on the
entire fid list and mark all related fids as unreclaimable (the reclaim
logic will just ignore them) and, of course, we open or re-open their
file descriptors if needed since we're about to unlink the file.
This is the purpose of v9fs_mark_fids_unreclaim(). Since the actual
opening of a file can cause the coroutine to yield, another client
request could possibly add a new fid that we may want to mark as
non-reclaimable as well. The loop is thus restarted if the re-open
request was actually transmitted to the backend. This is achieved
by keeping a reference on the first fid (head) before traversing
the list.
This is wrong in several ways:
- a potential clunk request from the client could tear the first
fid down and cause the reference to be stale. This leads to a
use-after-free error that can be detected with ASAN, using a
custom 9p client
- fids are added at the head of the list : restarting from the
previous head will always miss fids added by a some other
potential request
All these problems could be avoided if fids were being added at the
end of the list. This can be achieved with a QSIMPLEQ, but this is
probably too much change for a bug fix. For now let's keep it
simple and just restart the loop from the current head.
docs/system: Remove deprecated 'fulong2e' machine alias
The 'fulong2e' machine alias has been marked as deprecated since
QEMU v5.1 (commit c3a09ff68dd, the machine is renamed 'fuloong2e').
Time to remove it now.
target/mips: Remove vendor specific CPU definitions
Vendor specific CPU definitions are not very useful. Use the
ISA definitions instead, which are more helpful when looking
at the various CPU definitions.
nanoMIPS not a CPU, but an ISA. The nanoMIPS ISA is already
defined as ISA_NANOMIPS32.
Remove this incorrect definition and update the single CPU
implementing it, the I7200.
target/mips: Convert Rel6 LWLE/LWRE/SWLE/SWRE opcodes to decodetree
LWLE/LWRE/SWLE/SWRE (EVA) opcodes have been removed from
the Release 6. Add a single decodetree entry for the opcodes,
triggering Reserved Instruction if ever used.
target/mips: Introduce decodetree helpers for Release6 LSA/DLSA opcodes
LSA and LDSA opcodes are also available with MIPS release 6.
Introduce the decodetree config files and call the decode()
helpers in the main decode_opc() loop.
Extract 2200 lines from the huge translate.c to a new file,
'msa_translate.c'. As there are too many inter-dependencies
we don't compile it as another object yet, but keep including
it in the big translate.o. We gain in code maintainability.
target/mips: Extract msa_translate_init() from mips_tcg_init()
The msa_wr_d[] registers are only initialized/used by MSA.
They are declared static. We want to move them to the new
'msa_translate.c' unit in few commits, without having to
declare them global (with extern).
Extract first the logic initialization of the MSA registers
from the generic initialization. We will later move this
function along with the MSA registers to the new C unit.
target/mips: Alias MSA vector registers on FPU scalar registers
Commits 863f264d10f ("add msa_reset(), global msa register") and cb269f273fd ("fix multiple TCG registers covering same data")
removed the FPU scalar registers and replaced them by aliases to
the MSA vector registers.
It is not very clear to have FPU registers displayed with MSA
register names, even if MSA ASE is not present.
Instead of aliasing FPU registers to the MSA ones (even when MSA
is absent), we now alias the MSA ones to the FPU ones (only when
MSA is present).
Only decode MSA opcodes if MSA is present (implemented).
Now than check_msa_access() will only be called if MSA is
present, the only way to have MIPS_HFLAG_MSA unset is if
MSA is disabled (bit CP0C5_MSAEn cleared, see previous
commit). Therefore we can remove the 'reserved instruction'
exception.
target/mips/translate: Expose check_mips_64() to 32-bit mode
To allow compiling 64-bit specific translation code more
generically (and removing #ifdef'ry), allow compiling
check_mips_64() on 32-bit targets.
If ever called on 32-bit, we obviously emit a reserved
instruction exception.
target/mips/translate: Extract decode_opc_legacy() from decode_opc()
As we will slowly move to decodetree generated decoders,
extract the legacy decoding from decode_opc(), so new
decoders are added in decode_opc() while old code is
removed from decode_opc_legacy().
target/mips/translate: Add declarations for generic code
Some CPU translation functions / registers / macros and
definitions can be used by ISA / ASE / extensions out of
the big translate.c file. Declare them in "translate.h".
projects / qemu.git / log