| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
The JSON parser optionally supports interpolation. The lexer
recognizes interpolation tokens unconditionally. The parser rejects
them when interpolation is disabled, in parse_interpolation().
However, it neglects to set an error then, which can make
json_parser_parse() fail without setting an error.
Move the check for unwanted interpolation from the parser's
parse_interpolation() into the lexer's finite state machine. When
interpolation is disabled, '%' is now handled like any other
unexpected character.
The next commit will improve how such lexical errors are handled.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180823164025.12553-39-armbru@redhat.com>
|
|
The JSON parser optionally supports interpolation. The code calls it
"escape". Awkward, because it uses the same term for escape sequences
within strings. The latter usage is consistent with RFC 8259 "The
JavaScript Object Notation (JSON) Data Interchange Format" and ISO C.
Call the former "interpolation" instead.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180823164025.12553-38-armbru@redhat.com>
|
|
The classical way to structure parser and lexer is to have the client
call the parser to get an abstract syntax tree, the parser call the
lexer to get the next token, and the lexer call some function to get
input characters.
Another way to structure them would be to have the client feed
characters to the lexer, the lexer feed tokens to the parser, and the
parser feed abstract syntax trees to some callback provided by the
client. This way is more easily integrated into an event loop that
dispatches input characters as they arrive.
Our JSON parser is kind of between the two. The lexer feeds tokens to
a "streamer" instead of a real parser. The streamer accumulates
tokens until it got the sequence of tokens that comprise a single JSON
value (it counts curly braces and square brackets to decide). It
feeds those token sequences to a callback provided by the client. The
callback passes each token sequence to the parser, and gets back an
abstract syntax tree.
I figure it was done that way to make a straightforward recursive
descent parser possible. "Get next token" becomes "pop the first
token off the token sequence". Drawback: we need to store a complete
token sequence. Each token eats 13 + input characters + malloc
overhead bytes.
Observations:
1. This is not the only way to use recursive descent. If we replaced
"get next token" by a coroutine yield, we could do without a
streamer.
2. The lexer reports errors by passing a JSON_ERROR token to the
streamer. This communicates the offending input characters and
their location, but no more.
3. The streamer reports errors by passing a null token sequence to the
callback. The (already poor) lexical error information is thrown
away.
4. Having the callback receive a token sequence duplicates the code to
convert token sequence to abstract syntax tree in every callback.
5. Known bug: the streamer silently drops incomplete token sequences.
This commit rectifies 4. by lifting the call of the parser from the
callbacks into the streamer. Later commits will address 3. and 5.
The lifting removes a bug from qjson.c's parse_json(): it passed a
pointer to a non-null Error * in certain cases, as demonstrated by
check-qjson.c.
json_parser_parse() is now unused. It's a stupid wrapper around
json_parser_parse_err(). Drop it, and rename json_parser_parse_err()
to json_parser_parse().
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180823164025.12553-35-armbru@redhat.com>
|
|
json_lexer_init() takes the function to process a token as an
argument. It's always json_message_process_token(). Makes the code
harder to understand for no actual gain. Drop the indirection.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180823164025.12553-34-armbru@redhat.com>
|
|
The lexer always returns 0 when char feeding. Furthermore, none of the
caller care about the return value.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20180326150916.9602-10-marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180823164025.12553-32-armbru@redhat.com>
|
|
We reject bytes that can't occur in valid UTF-8 (\xC0..\xC1,
\xF5..\xFF in the lexer. That's insufficient; there's plenty of
invalid UTF-8 not containing these bytes, as demonstrated by
check-qjson:
* Malformed sequences
- Unexpected continuation bytes
- Missing continuation bytes after start bytes other than
\xC0..\xC1, \xF5..\xFD.
* Overlong sequences with start bytes other than \xC0..\xC1,
\xF5..\xFD.
* Invalid code points
Fixing this in the lexer would be bothersome. Fixing it in the parser
is straightforward, so do that.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180823164025.12553-23-armbru@redhat.com>
|
|
staging
check/next for 20180822
# gpg: Signature made Wed 22 Aug 2018 09:03:40 BST
# gpg: using RSA key F487EF185872D723
# gpg: Good signature from "Juan Quintela <quintela@redhat.com>"
# gpg: aka "Juan Quintela <quintela@trasno.org>"
# Primary key fingerprint: 1899 FF8E DEBF 58CC EE03 4B82 F487 EF18 5872 D723
* remotes/juanquintela/tags/check/20180822:
check: Only test tpm devices when they are compiled in
check: Only test usb-ehci when it is compiled in
check: Only test usb-uhci devices when they are compiled in
check: Only test usb-ohci when it is compiled in
check: Only test nvme when it is compiled in
check: Only test pvpanic when it is compiled in
check: Only test wdt_ib700 when it is compiled in
check: Only test sdhci when it is compiled in
check: Only test i82801b11 when it is compiled in
check: Only test ioh3420 when it is compiled in
check: Only test ipack when it is compiled in
check: Only test hda when it is compiled in
check: Only test ac97 when it is compiled in
check: Only test es1370 when it is compiled in
check: Only test rtl8139 when it is compiled in
check: Only test pcnet when it is compiled in
check: Only test eepro100 when it is compiled in
check: Only test ne2000 when it is compiled in
check: Only test vmxnet3 when it is compiled in
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
|
|
Validate the config settings that the guest tries to set.
The wiki page documentation is not really accurate here:
generally rather than failing requests to set bad parameters,
the hardware will just clip them to something sensible.
Validate the most important parameters: sizes and
the viewport offsets. This prevents the framebuffer
code from trying to read out-of-range memory.
In the property handling code, we validate the new parameters every
time we encounter a tag that sets them. This means we validate the
config multiple times if the request includes multiple config-setting
tags, but the code would require significant restructuring to do a
validation only once but still return the clipped settings for
get-parameter tags and the buffer allocation tag.
Validation of settings made via the older bcm2835_fb_mbox_push()
function will be done in the next commit.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180814144436.679-8-peter.maydell@linaro.org
|
|
The raspi framebuffir in bcm2835_fb supports the definition
of a virtual "viewport", which is smaller than the full
physical framebuffer size and at an adjustable offset within
it. Only the viewport area is sent to the screen. This allows
the guest to do things like double buffering, or scrolling
by adjusting the viewport origin. Currently QEMU doesn't
implement this at all.
Add support for this feature:
* the property mailbox code needs to distinguish the
virtual width/height from the physical width/height
* the framebuffer code needs to do something with the
virtual width/height/origin information
Note that the wiki documentation on the semantics of the
virtual and physical height and width has it the wrong way
around -- the virtual size is the size of the allocated
buffer, and the physical size is the size of the display,
so the virtual size is always the same as or larger than
the physical.
If the viewport size is set smaller than the physical
screen size, we ignore the viewport settings completely
and just display the physical screen area.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180814144436.679-7-peter.maydell@linaro.org
|
|
Abstract out the calculation of the pitch and size of the
framebuffer into functions that operate on the BCM2835FBConfig
struct -- these are about to get a little more complicated
when we add support for virtual and physical sizes differing.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180814144436.679-6-peter.maydell@linaro.org
|
|
The bcm2835_fb's initial resolution and other parameters are set
via QOM properties. We should reset to those initial values on
device reset, which means we need to save the QOM property
values somewhere that they are not overwritten by guest
changes to the framebuffer configuration.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180814144436.679-5-peter.maydell@linaro.org
|
|
The BCM2835FBState struct has a 'pitch' field which is a
cached copy of xres * (bpp >> 3), and a 'size' field which is
a cached copy of pitch * yres. However we don't actually do
anything with these fields; delete them. We retain the
now-unused slots in the VMState struct for migration
compatibility.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180814144436.679-4-peter.maydell@linaro.org
|
|
Refactor the fb property setting code so that rather than
using a set of pointers to local variables to track
whether a config value has been updated in the current
mbox and if so what its new value is, we just copy
all the current settings of the fb at the start, and
then update that copy as we go along, before asking
the fb to switch to it at the end.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180814144436.679-3-peter.maydell@linaro.org
|
|
The handling of framebuffer properties in the bcm2835_property code
is a bit clumsy, because for each of the many fb related properties
we try to track the value we're about to set and whether we're going
to be setting a value, and then we hand all the new values off
to the framebuffer via a function which takes them all as separate
arguments. It would be simpler if the property code could easily
copy all the framebuffer's current settings, update them with
the new specified values and then ask the framebuffer to switch
to the new set.
As the first part of this refactoring, pull all the fb config
settings fields in BCM2835FBState out into their own struct.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180814144436.679-2-peter.maydell@linaro.org
|
|
Create a new include file for the pl022's device struct,
type macros, etc, so that it can be instantiated using
the "embedded struct" coding style.
While we're adding the new file to MAINTAINERS, add
also the .c file, which was missing an entry.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 20180820141116.9118-16-peter.maydell@linaro.org
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
|
|
The IoTKit doesn't have any MSCs itself but it does need
some wiring to connect the external signals from MSCs
in the outer board model up to the registers and the
NVIC IRQ line.
We also need to expose a MemoryRegion corresponding to
the AHB bus, so that MSCs in the outer board model can
use that as their downstream port. (In the FPGA this is
the "AHB Slave Expansion" ports shown in the block
diagram in the AN505 documentation.)
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 20180820141116.9118-14-peter.maydell@linaro.org
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
|
|
The IoTKit does not have any Master Security Contollers itself,
but it does provide registers in the secure privilege control
block which allow control of MSCs in the external system.
Add support for these registers.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 20180820141116.9118-13-peter.maydell@linaro.org
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
|
|
Implement a model of the TrustZone Master Securtiy Controller,
as documented in the Arm CoreLink SIE-200 System IP for
Embedded TRM (DDI0571G):
https://developer.arm.com/products/architecture/m-profile/docs/ddi0571/g
The MSC is intended to sit in front of a device which can
be a bus master (eg a DMA controller) and programmably gate
its transactions. This allows a bus-mastering device to be
controlled by non-secure code but still restricted from
making accesses to addresses which are secure-only.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20180820141116.9118-12-peter.maydell@linaro.org
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
|
|
Wire up the system control element's register banks
(sysctl and sysinfo).
This is the last of the previously completely unimplemented
components in the IoTKit.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180820141116.9118-11-peter.maydell@linaro.org
|
|
Implement the IoTKit system control element's system information
block; this is just a pair of read-only version/config registers,
plus the usual PID/CID ID registers.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180820141116.9118-10-peter.maydell@linaro.org
|
|
The Arm IoTKit includes a system control element which
provides a block of read-only ID registers and a block
of read-write control registers. Implement a minimal
version of this.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180820141116.9118-9-peter.maydell@linaro.org
|
|
The IoTKit has a CMSDK timer device that runs on the S32KCLK.
Create this and wire it up.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180820141116.9118-8-peter.maydell@linaro.org
|
|
The IoTKit includes three different instances of the
CMSDK APB watchdog; create and wire them up.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180820141116.9118-7-peter.maydell@linaro.org
|
|
Now we have a model of the CMSDK dual timer, we can wire it
up in the IoTKit.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180820141116.9118-5-peter.maydell@linaro.org
|
|
The Arm Cortex-M System Design Kit includes a "dual-input timer module"
which combines two programmable down-counters. Implement a model
of this device.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180820141116.9118-4-peter.maydell@linaro.org
|
|
In the MPS2 FPGAIO, PSCNTR is a free-running downcounter with
a reload value configured via the PRESCALE register, and
COUNTER counts up by 1 every time PSCNTR reaches zero.
Implement these counters.
We can just increment the counters migration subsection's
version ID because we only added it in the previous commit,
so no released QEMU versions will be using it.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180820141116.9118-3-peter.maydell@linaro.org
|
|
The MPS2 FPGAIO block includes some simple free-running counters.
Implement these.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180820141116.9118-2-peter.maydell@linaro.org
|
|
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180814002653.12828-3-richard.henderson@linaro.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
|
|
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180814002653.12828-2-richard.henderson@linaro.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
|
|
* x86 TCG fixes for 64-bit call gates (Andrew)
* qumu-guest-agent freeze-hook tweak (Christian)
* pm_smbus improvements (Corey)
* Move validation to pre_plug for pc-dimm (David)
* Fix memory leaks (Eduardo, Marc-André)
* synchronization profiler (Emilio)
* Convert the CPU list to RCU (Emilio)
* LSI support for PPR Extended Message (George)
* vhost-scsi support for protection information (Greg)
* Mark mptsas as a storage device in the help (Guenter)
* checkpatch tweak cherry-picked from Linux (me)
* Typos, cleanups and dead-code removal (Julia, Marc-André)
* qemu-pr-helper support for old libmultipath (Murilo)
* Annotate fallthroughs (me)
* MemoryRegionOps cleanup (me, Peter)
* Make s390 qtests independent from libqos, which doesn't actually support it (me)
* Make cpu_get_ticks independent from BQL (me)
* Introspection fixes (Thomas)
* Support QEMU_MODULE_DIR environment variable (ryang)
# gpg: Signature made Thu 23 Aug 2018 17:46:30 BST
# gpg: using RSA key BFFBD25F78C7AE83
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>"
# gpg: aka "Paolo Bonzini <pbonzini@redhat.com>"
# Primary key fingerprint: 46F5 9FBD 57D6 12E7 BFD4 E2F7 7E15 100C CD36 69B1
# Subkey fingerprint: F133 3857 4B66 2389 866C 7682 BFFB D25F 78C7 AE83
* remotes/bonzini/tags/for-upstream: (69 commits)
KVM: cleanup unnecessary #ifdef KVM_CAP_...
target/i386: update MPX flags when CPL changes
i2c: pm_smbus: Add the ability to force block transfer enable
i2c: pm_smbus: Don't delay host status register busy bit when interrupts are enabled
i2c: pm_smbus: Add interrupt handling
i2c: pm_smbus: Add block transfer capability
i2c: pm_smbus: Make the I2C block read command read-only
i2c: pm_smbus: Fix the semantics of block I2C transfers
i2c: pm_smbus: Clean up some style issues
pc-dimm: assign and verify the "addr" property during pre_plug
pc: drop memory region alignment check for 0
util/oslib-win32: indicate alignment for qemu_anon_ram_alloc()
pc-dimm: assign and verify the "slot" property during pre_plug
ipmi: Use proper struct reference for BT vmstate
vhost-scsi: expose 't10_pi' property for VIRTIO_SCSI_F_T10_PI
vhost-scsi: unify vhost-scsi get_features implementations
vhost-user-scsi: move host_features into VHostSCSICommon
cpus: allow cpu_get_ticks out of BQL
cpus: protect TimerState writes with a spinlock
seqlock: add QemuLockable support
...
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
|
|
The PIIX4 hardware has block transfer buffer always enabled in
the hardware, but the i801 does not. Add a parameter to pm_smbus_init
to force on the block transfer so the PIIX4 handler can enable this
by default, as it was disabled by default before.
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <1534796770-10295-9-git-send-email-minyard@acm.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
Add the necessary code so that interrupts actually work from
the pm_smbus device.
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <1534796770-10295-7-git-send-email-minyard@acm.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
There was no block transfer code in pm_smbus.c, and it is needed
for some devices. So add it.
This adds both byte-by-byte block transfers and buffered block
transfers.
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <1534796770-10295-5-git-send-email-minyard@acm.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
The I2C block transfer commands was not implemented correctly, it
read a length byte and such like it was an smbus transfer.
So fix the smbus_read_block() and smbus_write_block() functions
so they can properly handle I2C transfers, and normal SMBus
transfers (for upcoming changes). Pass in a transfer size and
a bool to know whether to use the size byte (like SMBus) or use
the length given (like I2C).
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <1534796770-10295-3-git-send-email-minyard@acm.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
We can assign and verify the address before realizing and trying to plug.
reading/writing the address property should never fail for DIMMs, so let's
reduce error handling a bit by using &error_abort. Getting access to the
memory region now might however fail. So forward errors from
get_memory_region() properly.
As all memory devices should use the alignment of the underlying memory
region for guest physical address asignment, do detection of the
alignment in pc_dimm_pre_plug(), but allow pc.c to overwrite the
alignment for compatibility handling.
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180801133444.11269-5-david@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
We can assign and verify the slot before realizing and trying to plug.
reading/writing the slot property should never fail, so let's reduce
error handling a bit by using &error_abort.
To do this during pre_plug, add and use (x86, ppc) pc_dimm_pre_plug().
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180801133444.11269-2-david@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
In preparation for having vhost-scsi also make use of host_features,
move it from struct VHostUserSCSI into struct VHostSCSICommon.
Signed-off-by: Greg Edwards <gedwards@ddn.com>
Message-Id: <20180808195235.5843-2-gedwards@ddn.com>
Reviewed-by: Felipe Franciosi <felipe@nutanix.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
A shortcut when the seqlock write is protected by a spinlock or any mutex
other than the BQL.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
Iterating over the list without using atomics is undefined behaviour,
since the list can be modified concurrently by other threads (e.g.
every time a new thread is created in user-mode).
Fix it by implementing the CPU list as an RCU QTAILQ. This requires
a little bit of extra work to traverse list in reverse order (see
previous patch), but other than that the conversion is trivial.
Signed-off-by: Emilio G. Cota <cota@braap.org>
Message-Id: <20180819091335.22863-12-cota@braap.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
Signed-off-by: Emilio G. Cota <cota@braap.org>
Message-Id: <20180819091335.22863-5-cota@braap.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
Signed-off-by: Emilio G. Cota <cota@braap.org>
Message-Id: <20180819091335.22863-4-cota@braap.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
It's unnecessary because the pointer isn't dereferenced.
Signed-off-by: Emilio G. Cota <cota@braap.org>
Message-Id: <20180819091335.22863-3-cota@braap.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
To avoid undefined behaviour.
Signed-off-by: Emilio G. Cota <cota@braap.org>
Message-Id: <20180819091335.22863-2-cota@braap.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
Use kernel common header for fw_cfg.
(unfortunately, optionrom.h must have its own define, since it's
actually an assembler header)
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20180817155910.5722-2-marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
|
|
The BQL is acquired via qemu_mutex_lock_iothread(), which makes
the profiler assign the associated wait time (i.e. most of
BQL wait time) entirely to that function. This loses the original
call site information, which does not help diagnose BQL contention.
Fix it by tracking the callers explicitly.
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
I first implemented this by deleting all entries in the global
hash table. But doing that safely slows down profiling, since
we'd need to introduce rcu_read_lock/unlock in the fast path.
What's implemented here avoids messing with the thread-local
data in the global hash table. It achieves this by taking a snapshot
of the current state, so that subsequent reports present the delta
wrt to the snapshot.
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
The goal of this module is to profile synchronization primitives (i.e.
mutexes, recursive mutexes and condition variables) so that scalability
issues can be quickly diagnosed.
Sync primitives are profiled by QSP based on the vaddr of the object accessed
as well as the call site (file:line_nr). That means the same object called
from two different call sites will be tracked in separate entries, which
might be reported together or separately (see subsequent commit on
call site coalescing).
Some perf numbers:
Host: Intel(R) Core(TM) i7-6700K CPU @ 4.00GHz
Command: taskset -c 0 tests/atomic_add-bench -d 5 -m
- Before: 54.80 Mops/s
- After: 54.75 Mops/s
That is, a negligible slowdown due to the now indirect call to
qemu_mutex_lock. Note that using a branch instead of an indirect
call introduces a more severe slowdown (53.65 Mops/s, i.e. 2% slowdown).
Enabling the profiler (with -p, added in this series) is more interesting:
- No profiling: 54.75 Mops/s
- W/ profiling: 12.53 Mops/s
That is, a 4.36X slowdown.
We can break down this slowdown by removing the get_clock calls or
the entry lookup:
- No profiling: 54.75 Mops/s
- W/o get_clock: 25.37 Mops/s
- W/o entry lookup: 19.30 Mops/s
- W/ profiling: 12.53 Mops/s
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
