spapr_drc.c: use DRC reconfiguration to cleanup DIMM unplug state
Handling errors in memory hotunplug in the pSeries machine is more
complex than any other device type, because there are all the
complications that other devices has, and more.
For instance, determining a timeout for a DIMM hotunplug must consider
if it's a Hash-MMU or a Radix-MMU guest, because Hash guests takes
longer to hotunplug DIMMs. The size of the DIMM is also a factor, given
that longer DIMMs naturally takes longer to be hotunplugged from the
kernel. And there's also the guest memory usage to be considered: if
there's a process that is consuming memory that would be lost by the
DIMM unplug, the kernel will postpone the unplug process until the
process finishes, and then initiate the regular hotunplug process. The
first two considerations are manageable, but the last one is a deal
breaker.
There is no sane way for the pSeries machine to determine the memory
load in the guest when attempting a DIMM hotunplug - and even if there
was a way, the guest can start using all the RAM in the middle of the
unplug process and invalidate our previous assumptions - and in result
we can't even begin to calculate a timeout for the operation. This means
that we can't implement a viable timeout mechanism for memory unplug in
pSeries.
Going back to why we would consider an unplug timeout, the reason is
that we can't know if the kernel is giving up the unplug. Turns out
that, sometimes, we can. Consider a failed memory hotunplug attempt
where the kernel will error out with the following message:
'pseries-hotplug-mem: Memory indexed-count-remove failed, adding any
removed LMBs'
This happens when there is a LMB that the kernel gave up in removing,
and the LMBs previously marked for removal are now being added back.
This happens in the pseries kernel in [1], dlpar_memory_remove_by_ic()
into dlpar_add_lmb(), and after that update_lmb_associativity_index().
In this function, the kernel is configuring the LMB DRC connector again.
Note that this is a valid usage in LOPAR, as stated in section
"ibm,configure-connector RTAS Call":
'A subsequent sequence of calls to ibm,configure-connector with the same
entry from the “ibm,drc-indexes” or “ibm,drc-info” property will restart
the configuration of devices which were not completely configured.'
We can use this kernel behavior in our favor. If a DRC connector
reconfiguration for a LMB that we marked as unplug pending happens, this
indicates that the kernel changed its mind about the unplug and is
reasserting that it will keep using all the LMBs of the DIMM. In this
case, it's safe to assume that the whole DIMM device unplug was
cancelled.
This patch hops into rtas_ibm_configure_connector() and, in the scenario
described above, clear the unplug state for the DIMM device. This will
not solve all the problems we still have with memory unplug, but it will
cover this case where the kernel reconfigures LMBs after a failed
unplug. We are a bit more resilient, without using an unreliable
timeout, and we didn't make the remaining error cases any worse.
There is a reliable way to make a CPU hotunplug fail in the pseries
machine. Hotplug a CPU A, then offline all other CPUs inside the guest
but A. When trying to hotunplug A the guest kernel will refuse to do it,
because A is now the last online CPU of the guest. PAPR has no 'error
callback' in this situation to report back to the platform, so the guest
kernel will deny the unplug in silent and QEMU will never know what
happened. The unplug pending state of A will remain until the guest is
shutdown or rebooted.
Previous attempts of fixing it (see [1] and [2]) were aimed at trying to
mitigate the effects of the problem. In [1] we were trying to guess
which guest CPUs were online to forbid hotunplug of the last online CPU
in the QEMU layer, avoiding the scenario described above because QEMU is
now failing in behalf of the guest. This is not robust because the last
online CPU of the guest can change while we're in the middle of the
unplug process, and our initial assumptions are now invalid. In [2] we
were accepting that our unplug process is uncertain and the user should
be allowed to spam the IRQ hotunplug queue of the guest in case the CPU
hotunplug fails.
This patch presents another alternative, using the timeout
infrastructure introduced in the previous patch. CPU hotunplugs in the
pSeries machine will now timeout after 15 seconds. This is a long time
for a single CPU unplug to occur, regardless of guest load - although
the user is *strongly* encouraged to *not* hotunplug devices from a
guest under high load - and we can be sure that something went wrong if
it takes longer than that for the guest to release the CPU (the same
can't be said about memory hotunplug - more on that in the next patch).
Timing out the unplug operation will reset the unplug state of the CPU
and allow the user to try it again, regardless of the error situation
that prevented the hotunplug to occur. Of all the not so pretty
fixes/mitigations for CPU hotunplug errors in pSeries, timing out the
operation is an admission that we have no control in the process, and
must assume the worst case if the operation doesn't succeed in a
sensible time frame.
The LoPAR spec provides no way for the guest kernel to report failure of
hotplug/hotunplug events. This wouldn't be bad if those operations were
granted to always succeed, but that's far for the reality.
What ends up happening is that, in the case of a failed hotunplug,
regardless of whether it was a QEMU error or a guest misbehavior, the
pSeries machine is retaining the unplug state of the device in the
running guest. This state is cleanup in machine reset, where it is
assumed that this state represents a device that is pending unplug, and
the device is hotunpluged from the board. Until the reset occurs, any
hotunplug operation of the same device is forbid because there is a
pending unplug state.
This behavior has at least one undesirable side effect. A long standing
pending unplug state is, more often than not, the result of a hotunplug
error. The user had to dealt with it, since retrying to unplug the
device is noy allowed, and then in the machine reset we're removing the
device from the guest. This means that we're failing the user twice -
failed to hotunplug when asked, then hotunplugged without notice.
Solutions to this problem range between trying to predict when the
hotunplug will fail and forbid the operation from the QEMU layer, from
opening up the IRQ queue to allow for multiple hotunplug attempts, from
telling the users to 'reboot the machine if something goes wrong'. The
first solution is flawed because we can't fully predict guest behavior
from QEMU, the second solution is a trial and error remediation that
counts on a hope that the unplug will eventually succeed, and the third
is ... well.
This patch introduces a crude, but effective solution to hotunplug
errors in the pSeries machine. For each unplug done, we'll timeout after
some time. If a certain amount of time passes, we'll cleanup the
hotunplug state from the machine. During the timeout period, any unplug
operations in the same device will still be blocked. After that, we'll
assume that the guest failed the operation, and allow the user to try
again. If the timeout is too short we'll prevent legitimate hotunplug
situations to occur, so we'll need to overestimate the regular time an
unplug operation takes to succeed to account that.
The true solution for the hotunplug errors in the pSeries machines is a
PAPR change to allow for the guest to warn the platform about it. For
now, the work done in this timeout design can be used for the new PAPR
'abort hcall' in the future, given that for both cases we'll need code
to cleanup the existing unplug states of the DRCs.
At this moment we're adding the basic wiring of the timer into the DRC.
Next patch will use the timer to timeout failed CPU hotunplugs.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20210222194531.62717-4-danielhb413@gmail.com> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Fabiano Rosas [Mon, 22 Feb 2021 19:40:35 +0000 (16:40 -0300)]
target/ppc: Fix bcdsub. emulation when result overflows
The commit d03b174a83 (target/ppc: simplify bcdadd/sub functions)
meant to simplify some of the code but it inadvertently altered the
way the CR6 field is set after the operation has overflowed.
The CR6 bits are set based on the *unbounded* result of the operation,
so we need to look at the result before returning from bcd_add_mag,
otherwise we will look at 0 when it overflows.
The Power ISA 2.07B says:
If the unbounded result is greater than zero, do the following.
If PS=0, the sign code of the result is set to 0b1100.
If PS=1, the sign code of the result is set to 0b1111.
If the operation overflows, CR field 6 is set to 0b0101. Otherwise,
CR field 6 is set to 0b0100.
spapr: rename spapr_drc_detach() to spapr_drc_unplug_request()
spapr_drc_detach() is not the best name for what the function does. The
function does not detach the DRC, it makes an uncommited attempt to do
it. It'll mark the DRC as pending unplug, via the 'unplug_request'
flag, and only if the DRC state is drck->empty_state it will detach the
DRC, via spapr_drc_release().
This is a contrast with its pair spapr_drc_attach(), where the function
is indeed creating the DRC QOM object. If you know what
spapr_drc_attach() does, you can be misled into thinking that
spapr_drc_detach() is removing the DRC from QEMU internal state, which
isn't true.
The current role of this function is better described as a request for
detach, since there's no guarantee that we're going to detach the DRC in
the end. Rename the function to spapr_drc_unplug_request to reflect
what is is doing.
The initial idea was to change the name to spapr_drc_detach_request(),
and later on change the unplug_request flag to detach_request. However,
unplug_request is a migratable boolean for a long time now and renaming
it is not worth the trouble. spapr_drc_unplug_request() setting
drc->unplug_request is more natural than spapr_drc_detach_request
setting drc->unplug_request.
Reviewed-by: Greg Kurz <groug@kaod.org> Reviewed-by: David Gibson <david@gibson.dropbear.id.au> Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20210222194531.62717-3-danielhb413@gmail.com> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
spapr_drc.c: use spapr_drc_release() in isolate_physical/set_unusable
When moving a physical DRC to "Available", drc_isolate_physical() will
move the DRC state to STATE_PHYSICAL_POWERON and, if the DRC is marked
for unplug, call spapr_drc_detach(). For physical DRCs,
drck->empty_state is STATE_PHYSICAL_POWERON, meaning that we're sure
that spapr_drc_detach() will end up calling spapr_drc_release() in the
end.
Likewise, for logical DRCs, drc_set_unusable will move the DRC to
"Unusable" state, setting drc->state to STATE_LOGICAL_UNUSABLE, which is
the drck->empty_state for logical DRCs. spapr_drc_detach() will call
spapr_drc_release() in this case as well.
In both scenarios, spapr_drc_detach() is being used as a
spapr_drc_release(), wrapper, where we also set unplug_requested (which
is already true, otherwise spapr_drc_detach() wouldn't be called in the
first place) and check if drc->state == drck->empty_state, which we also
know it's guaranteed to be true because we just set it.
Just use spapr_drc_release() in these functions to be clear of our
intentions in both these functions.
Reviewed-by: Greg Kurz <groug@kaod.org> Reviewed-by: David Gibson <david@gibson.dropbear.id.au> Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20210222194531.62717-2-danielhb413@gmail.com> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This is mostly compiler warnings fixed but while doing this,
a bug in MIN() in tcgbios was found.
Alexey Kardashevskiy (14):
helpers: Define MIN()
libc: Compile with -Wextra
elf: Compile with -Wextra
usb: Compile with -Wextra
veth: Compile with -Wextra
virtio: Compile with -Wextra
e1000: Compile with -Wextra
libnet: Compile with -Wextra
libhv: Compile with -Wextra
libnvram: Compile with -Wextra
libtpm: Compile with -Wextra
slof/prim: Compile with -Wextra
Makefile: Actually compile with -Wextra
version: update to 20210217
Thomas Huth (1):
virtio-serial: Remove superfluous serial-* words
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
spapr_drc.c: do not call spapr_drc_detach() in drc_isolate_logical()
drc_isolate_logical() is used to move the DRC from the "Configured" to
the "Available" state, erroring out if the DRC is in the unexpected
"Unisolate" state and doing nothing (with RTAS_OUT_SUCCESS) if the DRC
is already in "Available" or in "Unusable" state.
When moving from "Configured" to "Available", the DRC is moved to the
LOGICAL_AVAILABLE state, a drc->unplug_requested check is done and, if
true, spapr_drc_detach() is called.
What spapr_drc_detach() does then is:
- set drc->unplug_requested to true. In fact, this is the only place
where unplug_request is set to true;
- does nothing else if drc->state != drck->empty_state. If the DRC
state is equal to drck->empty_state, spapr_drc_release() is
called. For logical DRCs, drck->empty_state = LOGICAL_UNUSABLE.
In short, calling spapr_drc_detach() in drc_isolate_logical() does
nothing. It'll set unplug_request to true again ('again' since it was
already true - otherwise the function wouldn't be called), and will
return without calling spapr_drc_release() because the DRC is not in
LOGICAL_UNUSABLE, since drc_isolate_logical() just moved it to
LOGICAL_AVAILABLE. The only place where the logical DRC is released is
when called from drc_set_unusable(), when it is moved to the
"Unusable" state. As it should, according to PAPR.
Even though calling spapr_drc_detach() in drc_isolate_logical() is
benign, removing it will avoid further thought about the matter. So
let's go ahead and do that.
As a note, this logic was introduced in commit bbf5c878ab76. Since
then, the DRC handling code was refactored and enhanced, and PAPR
itself went through some changes in the DRC area as well. It is
expected that some assumptions we had back then are now deprecated.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20210211225246.17315-2-danielhb413@gmail.com> Reviewed-by: Greg Kurz <groug@kaod.org> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Peter Maydell [Fri, 12 Feb 2021 18:06:52 +0000 (18:06 +0000)]
hw/display/sm501: Expand out macros in template header
Now that we only include sm501_template.h for the DEPTH==32 case, we
can expand out the uses of the BPP, PIXEL_TYPE and PIXEL_NAME macros
in that header.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20210212180653.27588-3-peter.maydell@linaro.org> Acked-by: BALATON Zoltan <balaton@eik.bme.hu> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Peter Maydell [Fri, 12 Feb 2021 18:06:51 +0000 (18:06 +0000)]
hw/display/sm501: Remove dead code for non-32-bit RGB surfaces
For a long time now the UI layer has guaranteed that the console
surface is always 32 bits per pixel RGB. Remove the legacy dead
code from the sm501 display device which was handling the
possibility that the console surface was some other format.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20210212180653.27588-2-peter.maydell@linaro.org> Acked-by: BALATON Zoltan <balaton@eik.bme.hu> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Peter Maydell [Tue, 9 Mar 2021 13:50:35 +0000 (13:50 +0000)]
Merge remote-tracking branch 'remotes/mcayland/tags/qemu-sparc-20210307' into staging
qemu-sparc queue
# gpg: Signature made Sun 07 Mar 2021 12:07:13 GMT
# gpg: using RSA key CC621AB98E82200D915CC9C45BC2C56FAE0F321F
# gpg: issuer "mark.cave-ayland@ilande.co.uk"
# gpg: Good signature from "Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>" [full]
# Primary key fingerprint: CC62 1AB9 8E82 200D 915C C9C4 5BC2 C56F AE0F 321F
* remotes/mcayland/tags/qemu-sparc-20210307: (42 commits)
esp: add support for unaligned accesses
esp: implement non-DMA transfers in PDMA mode
esp: add trivial implementation of the ESP_RFLAGS register
esp: convert cmdbuf from array to Fifo8
esp: convert ti_buf from array to Fifo8
esp: transition to message out phase after SATN and stop command
esp: add maxlen parameter to get_cmd()
esp: raise interrupt after every non-DMA byte transferred to the FIFO
esp: remove old deferred command completion mechanism
esp: defer command completion interrupt on incoming data transfers
esp: latch individual bits in ESP_RINTR register
esp: implement FIFO flush command
esp: add 4 byte PDMA read and write transfers
esp: remove pdma_origin from ESPState
esp: use FIFO for PDMA transfers between initiator and device
esp: fix PDMA target selection
esp: rename get_cmd_cb() to esp_select()
esp: remove CMD pdma_origin
esp: use in-built TC to determine PDMA transfer length
esp: use ti_wptr/ti_rptr to manage the current FIFO position for PDMA
...
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
* remotes/bonzini-gitlab/tags/for-upstream: (23 commits)
meson: Stop if cfi is enabled with system slirp
trace: skip qemu_set_log_filename if no "-D" option was passed
trace: fix "-trace file=..."
meson: adjust timeouts for some slower tests
build-sys: invoke ninja with -d keepdepfile
qemu-option: do not suggest using the delay option
scsi: move host_status handling into SCSI drivers
scsi: inline sg_io_sense_from_errno() into the callers.
scsi-generic: do not snoop the output of failed commands
scsi: Add mapping for generic SCSI_HOST status to sense codes
scsi: Rename linux-specific SG_ERR codes to generic SCSI_HOST error codes
qemu-config: add error propagation to qemu_config_parse
x86/pvh: extract only 4 bytes of start address for 32 bit kernels
elf_ops: correct loading of 32 bit PVH kernel
lsilogic: Use PCIDevice::exit instead of DeviceState::unrealize
accel: kvm: Add aligment assert for kvm_log_clear_one_slot
accel: kvm: Fix memory waste under mismatch page size
vl.c: do not execute trace_init_backends() before daemonizing
qom: Check for wellformed id in user_creatable_add_type()
chardev: add nodelay option
...
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Peter Maydell [Mon, 8 Mar 2021 11:57:36 +0000 (11:57 +0000)]
Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20210308' into staging
target-arm queue:
* sbsa-ref: remove cortex-a53 from list of supported cpus
* sbsa-ref: add 'max' to list of allowed cpus
* target/arm: Add support for FEAT_SSBS, Speculative Store Bypass Safe
* npcm7xx: add EMC model
* xlnx-zynqmp: Remove obsolete 'has_rpu' property
* target/arm: Speed up aarch64 TBL/TBX
* virtio-mmio: improve virtio-mmio get_dev_path alog
* target/arm: Use TCF0 and TFSRE0 for unprivileged tag checks
* target/arm: Restrict v8M IDAU to TCG
* target/arm/cpu: Update coding style to make checkpatch.pl happy
* musicpal, tc6393xb, omap_lcdc, tcx: drop dead code for non-32-bit-RGB surfaces
* Add new board: mps3-an524
* remotes/pmaydell/tags/pull-target-arm-20210308: (49 commits)
hw/arm/mps2: Update old infocenter.arm.com URLs
docs/system/arm/mps2.rst: Document the new mps3-an524 board
hw/arm/mps2-tz: Provide PL031 RTC on mps3-an524
hw/arm/mps2-tz: Stub out USB controller for mps3-an524
hw/arm/mps2-tz: Add new mps3-an524 board
hw/arm/mps2-tz: Get armv7m_load_kernel() size argument from RAMInfo
hw/arm/mps2-tz: Support ROMs as well as RAMs
hw/arm/mps2-tz: Set MachineClass default_ram info from RAMInfo data
hw/arm/mps2-tz: Make RAM arrangement board-specific
hw/arm/mps2-tz: Allow boards to have different PPCInfo data
hw/arm/mps2-tz: Size the uart-irq-orgate based on the number of UARTs
hw/arm/mps2-tz: Move device IRQ info to data structures
hw/arm/mps2-tz: Allow PPCPortInfo structures to specify device interrupts
hw/arm/mps2-tz: Correct wrong interrupt numbers for DMA and SPI
hw/misc/mps2-scc: Implement CFG_REG5 and CFG_REG6 for MPS3 AN524
hw/arm/mps2-tz: Make number of IRQs board-specific
hw/arm/mps2-tz: Condition IRQ splitting on number of CPUs, not board type
hw/arm/mps2-tz: Make FPGAIO switch and LED config per-board
hw/misc/mps2-fpgaio: Support SWITCH register
hw/misc/mps2-fpgaio: Make number of LEDs configurable by board
...
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Peter Maydell [Mon, 15 Feb 2021 11:51:38 +0000 (11:51 +0000)]
hw/arm/mps2: Update old infocenter.arm.com URLs
Update old infocenter.arm.com URLs to the equivalent developer.arm.com
ones (the old URLs should redirect, but we might as well avoid the
redirection notice, and the new URLs are pleasantly shorter).
This commit covers the links to the MPS2 board TRM, the various
Application Notes, the IoTKit and SSE-200 documents.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20210215115138.20465-25-peter.maydell@linaro.org
Peter Maydell [Mon, 15 Feb 2021 11:51:37 +0000 (11:51 +0000)]
docs/system/arm/mps2.rst: Document the new mps3-an524 board
Add brief documentation of the new mps3-an524 board.
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: 20210215115138.20465-24-peter.maydell@linaro.org
Peter Maydell [Mon, 15 Feb 2021 11:51:36 +0000 (11:51 +0000)]
hw/arm/mps2-tz: Provide PL031 RTC on mps3-an524
The AN524 has a PL031 RTC, which we have a model of; provide it
rather than an unimplemented-device stub.
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: 20210215115138.20465-23-peter.maydell@linaro.org
Peter Maydell [Mon, 15 Feb 2021 11:51:35 +0000 (11:51 +0000)]
hw/arm/mps2-tz: Stub out USB controller for mps3-an524
The AN524 has a USB controller (an ISP1763); we don't have a model of
it but we should provide a stub "unimplemented-device" for it. This
is slightly complicated because the USB controller shares a PPC port
with the ethernet controller.
Implement a make_* function which provides creates a container
MemoryRegion with both the ethernet controller and an
unimplemented-device stub for the USB controller.
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: 20210215115138.20465-22-peter.maydell@linaro.org
When the MacOS toolbox ROM transfers data from a target device to an unaligned
memory address, the first/last byte of a 16-bit transfer needs to be handled
separately. This means that the first byte is preloaded into the FIFO before
the transfer, or the last byte remains in the FIFO after the transfer.
The result of this is that the PDMA routines must be updated so that the FIFO
is loaded/unloaded if the last 16-bit word is used (rather than the last byte)
and any remaining byte from a FIFO wraparound is handled correctly.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-43-mark.cave-ayland@ilande.co.uk>
esp: add trivial implementation of the ESP_RFLAGS register
The bottom 5 bits contain the number of bytes remaining in the FIFO which is
trivial to implement with Fifo8 (the remaining bits are unimplemented and left
as 0 for now).
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-41-mark.cave-ayland@ilande.co.uk>
Rename ESP_CMDBUF_SZ to ESP_CMDFIFO_SZ and cmdbuf_cdb_offset to cmdfifo_cdb_offset
to indicate that the command buffer type has changed from an array to a Fifo8.
This also enables us to remove the ESPState field cmdlen since the command length
is now simply the number of elements used in cmdfifo.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-40-mark.cave-ayland@ilande.co.uk>
esp: transition to message out phase after SATN and stop command
The SCSI bus should remain in the message out phase after the SATN and stop
command rather than transitioning to the command phase. A new ESPState variable
cmdbuf_cdb_offset is added which stores the offset of the CDB from the start
of cmdbuf when accumulating extended message out phase data.
Currently any extended message out data is discarded in do_cmd() before the CDB
is processed in do_busid_cmd().
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-38-mark.cave-ayland@ilande.co.uk>
Some guests use a mixture of DMA and non-DMA transfers in combination with the
SATN and stop command to transfer message out phase and command phase bytes to
the target. Prepare for the next commit by adding a maxlen parameter to
get_cmd() to allow partial transfers.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-37-mark.cave-ayland@ilande.co.uk>
esp: remove old deferred command completion mechanism
Commit ea84a44250 "scsi: esp: Defer command completion until previous interrupts
have been handled" provided a mechanism to delay the command completion interrupt
until ESP_RINTR is read after the command has completed.
With the previous fixes for latching the ESP_RINTR bits and deferring the setting
of the command completion interrupt for incoming data to the SCSI callback, this
workaround is no longer required and can be removed.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-35-mark.cave-ayland@ilande.co.uk>
esp: defer command completion interrupt on incoming data transfers
The MacOS toolbox ROM issues a command to the ESP controller as part of its
"FAST" SCSI routines and then proceeds to read the incoming data soon after
receiving the command completion interrupt.
Unfortunately due to SCSI block transfers being asynchronous the incoming data
may not yet be present causing an underflow error. Resolve this by waiting for
the SCSI subsystem transfer_data callback before raising the command completion
interrupt.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-34-mark.cave-ayland@ilande.co.uk>
Currently the ESP_RINTR register is set to a specific value as required within
the ESP state machine. In order to implement the upcoming deferred interrupt
functionality it is necessary to set individual bits within ESP_RINTR so that
a deferred interrupt will not overwrite the value of any other interrupt bits.
This also requires fixing up a few locations where the ESP_RINTR and ESP_RSEQ
registers are set/reset unexpectedly.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-33-mark.cave-ayland@ilande.co.uk>
The MacOS toolbox ROM performs 4 byte reads/writes when transferring data to
and from the target. Since the SCSI bus is 16-bits wide, use the memory API
to split a 4 byte access into 2 x 2 byte accesses.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-31-mark.cave-ayland@ilande.co.uk>
Now that all data is transferred via the FIFO (ti_buf) there is no need to track
the source buffer being used for the data transfer. This also eliminates the
need for a separate subsection for PDMA state migration.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-30-mark.cave-ayland@ilande.co.uk>
esp: use FIFO for PDMA transfers between initiator and device
PDMA as implemented on the Quadra 800 uses DREQ to load data into the FIFO
up to a maximum of 16 bytes at a time. The MacOS toolbox ROM requires this
because it mixes FIFO and PDMA transfers whilst checking the FIFO status
and counter registers to ensure success.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-29-mark.cave-ayland@ilande.co.uk>
Currently the target selection for PDMA is done after the SCSI command has been
delivered which is not correct. Perform target selection as part of the initial
get_cmd() call when the command is submitted: if no target is present, don't
raise DRQ.
If the target is present then switch to the command phase since the MacOS toolbox
ROM checks for this before attempting to submit the SCSI command.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-28-mark.cave-ayland@ilande.co.uk>
The cmdbuf is really just a copy of FIFO data (including extra message phase
bytes) so its pdma_origin is effectively TI. Fortunately we already know when
we are receiving a SCSI command since do_cmd == 1 which enables us to
distinguish between the two cases in esp_pdma_read()/esp_pdma_write().
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-26-mark.cave-ayland@ilande.co.uk>
esp: use in-built TC to determine PDMA transfer length
Real hardware simply counts down using the in-built TC to determine when the
the PDMA request is complete. Use the TC to determine the PDMA transfer length
which then enables us to remove the redundant pdma_len variable.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-25-mark.cave-ayland@ilande.co.uk>
esp: move PDMA length adjustments into esp_pdma_read()/esp_pdma_write()
Here the updates to async_len and ti_size are moved into the corresponding
esp_pdma_read()/esp_pdma_write() function to eliminate the reference to
pdma_cur in do_dma_pdma_cb().
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-23-mark.cave-ayland@ilande.co.uk>
Now that PDMA SCSI commands are accumulated in cmdbuf in the same way as normal
commands, the existing logic for locating the start of the SCSI command in
cmdbuf via cmdlen can be used. This enables the PDMA-specific pdma_start and
also get_pdma_buf() to be removed.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-22-mark.cave-ayland@ilande.co.uk>
esp: remove the buf and buflen parameters from get_cmd()
Now that all SCSI commands are accumulated in cmdbuf, remove the buf and buflen
parameters from get_cmd() since these always reference cmdbuf and ESP_CMDBUF_SZ
respectively.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-21-mark.cave-ayland@ilande.co.uk>
esp: accumulate SCSI commands for PDMA transfers in cmdbuf instead of pdma_buf
ESP SCSI commands are already accumulated in cmdbuf and so there is no need to
keep a separate pdma_buf buffer. Accumulate SCSI commands for PDMA transfers in
cmdbuf instead of pdma_buf so update cmdlen accordingly and change pdma_origin
for PDMA transfers to CMD which allows the PDMA origin to be removed.
This commit also removes a stray memcpy() from get_cmd() which is a no-op because
cmdlen is always zero at the start of a command.
Notionally the removal of pdma_buf from vmstate_esp_pdma also breaks migration
compatibility for the PDMA subsection until its complete removal by the end of
the series.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-19-mark.cave-ayland@ilande.co.uk>
The ESP device already keeps track of the remaining bytes left to transfer via
its TC (transfer counter) register which is decremented for each byte that
is transferred across the SCSI bus.
Switch the transfer logic to use the value of TC instead of dma_left and then
remove dma_left completely, adding logic to the vmstate_esp post_load() function
to transfer the old dma_left value to the TC register during migration from
older versions.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-14-mark.cave-ayland@ilande.co.uk>
The value of dma_counter is set once at the start of the transfer and remains
the same until the transfer is complete. This allows the check in esp_transfer_data
to be simplified since dma_left will always be non-zero until the transfer is
completed.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-13-mark.cave-ayland@ilande.co.uk>
esp: apply transfer length adjustment when STC is zero at TC load time
Perform the length adjustment whereby a value of 0 in the STC represents
a transfer length of 0x10000 at the point where the TC is loaded at the
start of a DMA command rather than just when a TI (Transfer Information)
command is executed. This better matches the description as given in the
datasheet.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-12-mark.cave-ayland@ilande.co.uk>
esp: determine transfer direction directly from SCSI phase
The transfer direction is currently determined by checking the sign of ti_size
but as this series progresses ti_size can be zero at the end of the transfer.
Use the SCSI phase to determine the transfer direction as used in other SCSI
controller implementations.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-9-mark.cave-ayland@ilande.co.uk>
The QOM object representing ESPState is currently embedded within both the
SYSBUS_ESP and PCI_ESP devices with migration state handled by embedding
vmstate_esp within each device using VMSTATE_STRUCT.
Since the vmstate_esp fields are embedded directly within the migration
stream, the incoming vmstate_esp version_id is lost. The only version information
available is that from vmstate_sysbus_esp_scsi and vmstate_esp_pci_scsi, but
those versions represent their respective devices and not that of the underlying
ESPState.
Resolve this by adding a new version-dependent field in vmstate_sysbus_esp_scsi
and vmstate_esp_pci_scsi which stores the vmstate_esp version_id field within
ESPState to be used to allow migration from older QEMU versions.
Finally bump the vmstate_esp version to 5 to cover the upcoming ESPState changes
within this patch series.
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20210304221103.6369-5-mark.cave-ayland@ilande.co.uk>
Taylor Simpson [Thu, 4 Mar 2021 17:37:05 +0000 (09:37 -0800)]
target/hexagon: Fix shift amount check in fASHIFTL/fLSHIFTR
Fixes: a646e99cb90 ("Hexagon (target/hexagon) macros")
Eliminate the following Coverity CIDs (Bad bit shift operation)
325227
325292
325425
325526
325561
325564
325578
325637
325736
325748
325786
325815
325837
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Message-Id: <1614879425-9259-1-git-send-email-tsimpson@quicinc.com> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Commit 3e7a84eeccc ("Hexagon build infrastructure") added Hexagon
definitions that should be poisoned on target independent device
code, but forgot to update "exec/poison.h". Do it now.
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Taylor Simpson <tsimpson@quicinc.com>
Message-Id: <20210219135754.1968100-1-f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Alex Bennée [Wed, 24 Feb 2021 16:58:10 +0000 (16:58 +0000)]
include/exec: lightly re-arrange TranslationBlock
Lets make sure all the flags we compare when looking up blocks are
together in the same place.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20210224165811.11567-5-alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Alex Bennée [Wed, 24 Feb 2021 16:58:09 +0000 (16:58 +0000)]
accel/tcg: drop the use of CF_HASH_MASK and rename params
We don't really deal in cf_mask most of the time. The one time it's
relevant is when we want to remove an invalidated TB from the QHT
lookup. Everywhere else we should be looking up things without
CF_INVALID set.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20210224165811.11567-4-alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Alex Bennée [Wed, 24 Feb 2021 16:58:08 +0000 (16:58 +0000)]
accel/tcg: move CF_CLUSTER calculation to curr_cflags
There is nothing special about this compile flag that doesn't mean we
can't just compute it with curr_cflags() which we should be using when
building a new set.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20210224165811.11567-3-alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Alex Bennée [Wed, 24 Feb 2021 16:58:07 +0000 (16:58 +0000)]
accel/tcg: rename tb_lookup__cpu_state and hoist state extraction
Having a function return either and valid TB and some system state
seems excessive. It will make the subsequent re-factoring easier if we
lookup the current state where we are.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20210224165811.11567-2-alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
tcg/tci: Merge identical cases in generation (load/store opcodes)
Use CASE_32_64 and CASE_64 to reduce ifdefs and merge
cases that are identical between 32-bit and 64-bit hosts.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20210217202036.1724901-5-richard.henderson@linaro.org>
[PMD: Split patch as 5/5] Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20210218232840.1760806-6-f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
tcg/tci: Merge identical cases in generation (conditional opcodes)
Use CASE_32_64 and CASE_64 to reduce ifdefs and merge
cases that are identical between 32-bit and 64-bit hosts.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20210217202036.1724901-5-richard.henderson@linaro.org>
[PMD: Split patch as 4/5] Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20210218232840.1760806-5-f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
tcg/tci: Merge identical cases in generation (deposit opcode)
Use CASE_32_64 and CASE_64 to reduce ifdefs and merge
cases that are identical between 32-bit and 64-bit hosts.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20210217202036.1724901-5-richard.henderson@linaro.org>
[PMD: Split patch as 3/5] Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20210218232840.1760806-4-f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
tcg/tci: Merge identical cases in generation (exchange opcodes)
Use CASE_32_64 and CASE_64 to reduce ifdefs and merge
cases that are identical between 32-bit and 64-bit hosts.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20210217202036.1724901-5-richard.henderson@linaro.org>
[PMD: Split patch as 2/5] Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20210218232840.1760806-3-f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
tcg/tci: Merge identical cases in generation (arithmetic opcodes)
Use CASE_32_64 and CASE_64 to reduce ifdefs and merge
cases that are identical between 32-bit and 64-bit hosts.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20210217202036.1724901-5-richard.henderson@linaro.org>
[PMD: Split patch as 1/5] Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20210218232840.1760806-2-f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
tcg: Manage splitwx in tc_ptr_to_region_tree by hand
The use in tcg_tb_lookup is given a random pc that comes from the pc
of a signal handler. Do not assert that the pointer is already within
the code gen buffer at all, much less the writable mirror of it.
Fixes: db0c51a3803 Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Use the provided cpu_ldst.h interfaces. This fixes the build vs
the unconverted uses of g2h(), adds missed memory trace events,
and correctly recognizes when a SIGSEGV belongs to the guest via
set_helper_retaddr().
Fixes: 3e8f1628e864 Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
The ((bl < 0) ^ sub) condition which negates bl is incorrect and will
always lead to this abort. If the constant is positive, sub will make
it negative; if the constant is negative, sub will keep it negative.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
hw/sh4/sh7750_regs: Replace link to license by its full content
This file is borrowed from the RTEMS source code, which comes
with a GPL-2.0-or-later license with a header exception.
Expand the GPL-2.0-or-later license in place to not be dependent
on a 3rd party website. This also fix the misleading comment "The
license and distribution terms for this file may be found in the
file LICENSE in this distribution" referring to the RTEMS distribution
and not to the QEMU one.
Suggested-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20210222185605.2714192-1-f4bug@amsat.org>
As replaced the generic CONFIG_SH4 by more fine-grained
selectors, we can remove this now unused config variable.
Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210222141514.2646278-9-f4bug@amsat.org>
We want to be able to use the 'SH4' config for architecture
specific features. Add more fine-grained selection by adding
a CONFIG_SH_PCI selector for the SH4 PCI controller.
Move the file with the other PCI host devices in hw/pci-host
and add its missing MAINTAINERS entries.
Suggested-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210222141514.2646278-8-f4bug@amsat.org>
Add more fine-grained selection by adding a CONFIG_TC58128
selector for the TC58128 eeprom.
As this device is only used by the Shix machine, add an entry
to the proper section in MAINTAINERS.
Suggested-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210222141514.2646278-7-f4bug@amsat.org>
We want to be able to use the 'SH4' config for architecture
specific features. Add more fine-grained selection by adding
a CONFIG_SH_TIMER selector for the SH4 timer control unit.
Add the missing MAINTAINERS entries.
Suggested-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210222141514.2646278-6-f4bug@amsat.org>
We want to be able to use the 'SH4' config for architecture
specific features. Add more fine-grained selection by adding
a CONFIG_SH_SCI selector for the SH4 serial controller.
Add the missing MAINTAINERS entries.
Suggested-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20210222141514.2646278-5-f4bug@amsat.org>
projects / qemu-xen.git / log