[PATCH 76/88] esp.c: only transfer non-DMA COMMAND phase data for specific commands

The contents of the FIFO should only be copied to cmdfifo for ESP commands that are sending data to the SCSI bus, which are the SEL_* commands and the TI command. Otherwise any incoming data should be held in the FIFO as normal. This fixes booting of really old 32-bit SPARC Linux kernels such as A

[PATCH 25/88] esp.c: remove unaligned adjustment in do_dma_pdma_cb() to device path

With the latest changes this condition cannot occur anymore and so the logic can be completely removed. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 9 - 1 file changed, 9 deletions(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c index ec82097a01..3db90c9ab7 100644 --- a/hw/scsi/es

[PATCH 72/88] esp.c: handle TC underflow for DMA SCSI requests

Detect the case where the guest underflows TC by requesting a DMA transfer which is larger than the available data. If this case is detected, immediately complete the SCSI request and handle any remaining FIFO accesses in the STATUS phase by raising INTR_BS once the FIFO is below the threshold. No

[PATCH 22/88] esp.c: ensure that the PDMA callback is called for every device read

Rather than wait for the FIFO to fill up before calling the PDMA callback, push that logic directly into the from_device logic in do_dma_pdma_cb(). Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 8 +++- 1 file changed, 3 insertions(+), 5 deletions(-) diff --git a/hw/scsi/esp.c b/hw/scs

[PATCH 08/88] esp: move command execution logic to new esp_run_cmd() function

This helps to simplify esp_reg_write() and potentially allows for a 2-level deep FIFO to be implemented in future. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 177 ++ 1 file changed, 92 insertions(+), 85 deletions(-) diff --git a/hw/scsi/e

[PATCH 20/88] esp.c: update condition for esp_dma_done() in esp_do_dma() from device path

No change to the condition itself, other than to clarify that esp_dma_done() must be called when TC is zero. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 16 +--- 1 file changed, 5 insertions(+), 11 deletions(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c index dc515e6435..967

[PATCH 37/88] esp.c: introduce esp_get_phase() function

Make use of this new function in all places where the SCSI phase bits are manually masked from the ESP_RSTAT register. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 23 ++- 1 file changed, 14 insertions(+), 9 deletions(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c inde

[PATCH 70/88] esp.c: ensure that STAT_INT is cleared when reading ESP_RINTR

Both esp_raise_irq() and esp_lower_irq() check the STAT_INT bit in ESP_RSTAT to ensure that the IRQ is raised or lowered if its state changes. When reading ESP_RINTR, esp_lower_irq() was being called *after* ESP_RSTAT had been cleared meaning that STAT_INT was already clear, and so if STAT_INT was

[PATCH 53/88] esp.c: replace do_dma_pdma_cb() with esp_do_dma()

Now that the DMA logic is identical between do_dma_pdma_cb() and esp_do_dma() we can replace do_dma_pdma_cb() with esp_do_dma(). Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 138 +- 1 file changed, 1 insertion(+), 137 deletions(-) diff --gi

[PATCH 32/88] esp.c: copy PDMA logic for transfers to device from do_dma_pdma_cb() to esp_do_dma()

This is so that PDMA transfers can be performend by esp_do_dma() as well as do_dma_pdma_cb(). Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 9 + 1 file changed, 9 insertions(+) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c index 14c6d05253..65b4baab83 100644 --- a/hw/scsi/esp.c +++ b

[PATCH 43/88] esp.c: convert esp_do_nodma() do_cmd path to check for SCSI phase instead

Currently do_cmd is used to determine whether MESSAGE OUT and COMMAND phase data is being accumulated in cmdfifo. Update esp_do_dma() to check directly for these two SCSI phases instead. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 8 1 file changed, 4 insertions(+), 4 deletions(

[PATCH 57/88] esp.c: rename data_in_ready to to data_ready

This field is currently used to handle deferred interrupts for the DATA IN phase but the code will soon be updated to do the same for the DATA OUT phase. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 8 include/hw/scsi/esp.h | 2 +- 2 files changed, 5 insertions(+), 5 dele

[PATCH 41/88] esp.c: convert esp_do_dma() do_cmd path to check for SCSI phase instead

Currently do_cmd is used to determine whether MESSAGE OUT and COMMAND phase data is being accumulated in cmdfifo. Update esp_do_dma() to check directly for these two SCSI phases instead. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 13 + 1 file changed, 5 insertions(+), 8 dele

[PATCH 50/88] esp.c: move CMD_TI end of message phase detection to esp_do_dma() and do_dma_pdma_cb()

The existing check for TC == 0 is only valid during a TI command. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 32 ++-- 1 file changed, 18 insertions(+), 14 deletions(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c index f8c20d0584..9f787d12a8 100644 --- a/hw/sc

[PATCH 82/88] esp.c: consolidate DMA and PDMA logic in STATUS and MESSAGE IN phases

This allows the removal of duplicate logic shared between the two implementations. Note that we restrict esp_raise_drq() to PDMA to help reduce the log verbosity for normal DMA. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 7 +++ 1 file changed, 3 insertions(+), 4 deletions(-) diff

[PATCH 51/88] esp.c: don't use get_cmd() for CMD_SEL DMA commands

This can now be done using the existing logic in esp_do_dma() and do_dma_pdma_cb(). Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 22 ++ 1 file changed, 10 insertions(+), 12 deletions(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c index 9f787d12a8..3cf8b2b4eb 100644 --

[PATCH 42/88] esp.c: convert do_dma_pdma_cb() do_cmd path to check for SCSI phase instead

Currently do_cmd is used to determine whether MESSAGE OUT and COMMAND phase data is being accumulated in cmdfifo. Update esp_do_dma() to check directly for these two SCSI phases instead. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 8 1 file changed, 4 insertions(+), 4 deletions(

[PATCH 65/88] esp.c: move non-DMA TI logic to separate esp_nodma_ti_dataout() function

This is to allow the logic to be moved during the next commit. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 51 +-- 1 file changed, 29 insertions(+), 22 deletions(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c index dd6bf6f033..97e48e9526 10

[PATCH 85/88] esp.c: rename irq_data IRQ to drq_irq

The IRQ represented by irq_data is actually the DRQ (DMA request) line so rename it accordingly. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 8 include/hw/scsi/esp.h | 2 +- 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c inde

[PATCH 62/88] esp.c: zero command register when TI command terminates due to phase change

This is the behaviour documented in the datasheet and allows the state machine to correctly process multiple consecutive TI commands. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 8 1 file changed, 8 insertions(+) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c index 4c1ca63a57..ccb8

[PATCH 88/88] esp.c: add my copyright to the file

This series has involved rewriting and/or updating a considerable part of the ESP emulation so update the copyright in esp.c to reflect this. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 1 + 1 file changed, 1 insertion(+) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c index b8762d5ee0..590

[PATCH 74/88] esp.c: handle non-DMA FIFO writes used to terminate DMA commands

Certain versions of MacOS send the first 5 bytes of the CDB using DMA and then send the last byte of the CDB by writing to the FIFO. Update the non-DMA state machine to detect the end of the CDB and execute the SCSI command using similar logic as that which already exists for transferring the remai

[PATCH 60/88] esp.c: use deferred interrupts for both DATA IN and DATA OUT phases

This brings DATA OUT transfers in line with DATA IN transfers by ensuring that the guest visible function complete interrupt is only set once the SCSI layer has returned. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 35 ++- 1 file changed, 18 insertions(+),

[PATCH 81/88] esp.c: remove redundant n variable in PDMA COMMAND phase

This variable can be replaced by the existing len variable. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 8 +++- 1 file changed, 3 insertions(+), 5 deletions(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c index ae65c2ef37..8ef6d203e0 100644 --- a/hw/scsi/esp.c +++ b/hw/scsi/esp.c @@ -

[PATCH 39/88] esp.c: convert do_dma_pdma_db() to switch statement based upon SCSI phase

Currently only the DATA IN and DATA OUT phases are supported. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 19 --- 1 file changed, 12 insertions(+), 7 deletions(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c index 67d1d39db2..f6d05b0de7 100644 --- a/hw/scsi/esp.c +++ b/hw/

[PATCH 86/88] esp.c: keep track of the DRQ state during DMA

Currently the DRQ IRQ is updated every time DMA data is sent/received which is both inefficient and causes excessive logging of the DRQ state. Add a new drq_state bool that only updates the DRQ IRQ if its state changes. This commit adds the new drq_state bool to the migration state: since the vers

[PATCH 55/88] esp.c: always use esp_do_dma() in pdma_cb()

There is now only a single implementation contained within esp_do_dma() so call it directly. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 8 +--- 1 file changed, 1 insertion(+), 7 deletions(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c index c6e5ddd537..bdbdb209f7 100644 --- a/hw/scs

[PATCH 58/88] esp.c: separate logic based upon ESP command in esp_command_complete()

The handling of the INTR_FC and INTR_BS bits is different depending upon the last command executed by the ESP. Note that currently INTR_FC is managed elsewhere, but that will change soon. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 30 -- 1 file changed, 16 in

[PATCH 87/88] esp.c: switch TypeInfo registration to use DEFINE_TYPES() macro

The use of the DEFINE_TYPES() macro will soon be recommended over the use of calling type_init() directly. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 39 +-- 1 file changed, 17 insertions(+), 22 deletions(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c

[PATCH 67/88] esp.c: replace get_cmd() with esp_do_nodma()

Now that the esp_do_nodma() state machine correctly handles incoming FIFO data, all remaining users of get_cmd() can be replaced with esp_do_nodma() and the get_cmd() function removed completely. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 54 -

[PATCH 79/88] esp.c: consolidate DMA and PDMA logic in DATA IN phase

This allows the removal of duplicate logic shared between the two implementations. Note that we restrict esp_raise_drq() to PDMA to help reduce the log verbosity for normal DMA. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 51 +-- 1 file ch

[PATCH 31/88] esp.c: rename esp_dma_done() to esp_dma_ti_check()

This is because a single DMA request can be transferred using multiple TI commands, and so a TC equal to zero may not represent the completion of the SCSI DMA command. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 18 +- 1 file changed, 9 insertions(+), 9 deletions(-) diff

[PATCH 77/88] esp.c: only transfer non-DMA MESSAGE OUT phase data for specific commands

The contents of the FIFO should only be copied to cmdfifo for ESP commands that are sending data to the SCSI bus, which are the SEL_* commands and the TI command. Otherwise any incoming data should be held in the FIFO as normal. This fixes booting of NetBSD m68k under the Q800 machine once again.

[PATCH 80/88] esp.c: consolidate DMA and PDMA logic in MESSAGE OUT phase

This allows the removal of duplicate logic shared between the two implementations. Note that we restrict esp_raise_drq() to PDMA to help reduce the log verbosity for normal DMA. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 12 +--- 1 file changed, 5 insertions(+), 7 deletions(-)

[PATCH 19/88] esp: remove zero transfer size check from esp_do_dma()

The TI end of transfer interrupt only occurs when the TC reaches zero and is not related to the SCSI layer transfer. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c index 5b9c3f1e5e..dc515e6435 10

[PATCH 84/88] esp.c: implement DMA Transfer Pad command for DATA phases

The Transfer Pad command is used to either drop incoming FIFO data during the DATA IN phase or generate a series of zero bytes in the FIFO during the DATA OUT phase. Implement the DMA Transfer Pad command for the DATA phases which is used by the NeXTCube firmware in the DATA IN phase to ignore par

[PATCH 83/88] esp.c: replace n variable with len in esp_do_nodma()

This brings esp_do_nodma() in line with esp_do_dma(). Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 40 1 file changed, 20 insertions(+), 20 deletions(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c index 879e311bc4..73379a3c65 100644 --- a/hw/scsi/e

[PATCH 78/88] esp.c: consolidate DMA and PDMA logic in DATA OUT phase

This allows the removal of duplicate logic shared between the two implementations. Note that we restrict esp_raise_drq() to PDMA to help reduce the log verbosity for normal DMA. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 35 --- 1 file changed, 12 insert

[PATCH 73/88] esp.c: remove restriction on FIFO read access when DMA memory routines defined

The latest state machines can handle mixing DMA and non-DMA FIFO access for all SCSI phases except DATA IN and DATA OUT. For DATA IN and DATA OUT phases, the transfer is complete when TC == 0 and the updated logic will now handle TC underflow correctly, which makes it just about impossible to manua

[PATCH 75/88] esp.c: improve ESP_RSEQ logic consolidation

The ESP_RSEQ logic is scattered in a few places throughout the ESP state machine which is mainly because the ESP_RSEQ register isn't always reset when executing an ESP select command. Once this is done, the ESP_RSEQ register only needs to be updated at the point where the sequencer command complete

Re: [PULL 22/22] Revert "tests/avocado: remove skips from replay_kernel"

Peter Maydell writes: > On Fri, 12 Jan 2024 at 11:11, Alex Bennée wrote: >> >> This reverts commit c2ef5ee89d76f0ab77c4dd6a1c9eeed4d35d20ed. >> >> While the fixes for #2010 and #2013 have improved things locally it >> seems GitLab still continues to be flaky. >> >> Signed-off-by: Alex Bennée >>

[PATCH 66/88] esp.c: process non-DMA FIFO writes in esp_do_nodma()

Currently any write to the ESP FIFO in the MESSAGE OUT or COMMAND phases will manually raise the bus service interrupt. Instead of duplicating the interrupt logic in esp_reg_write(), update esp_do_nodma() to correctly process incoming FIFO data during the MESSAGE OUT and COMMAND phases. Part of thi

[PATCH 68/88] esp.c: move write_response() non-DMA logic to esp_do_nodma()

This moves the remaining non-DMA STATUS and MESSAGE IN phase logic from write_response() to esp_do_nodma(). Note that we can also now drop the extra fifo_reset() which is no longer required. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 32 +++- 1 file changed,

[PATCH 49/88] esp.c: move CMD_SELATN end of message phase detection to esp_do_dma() and do_dma_pdma_cb()

The special logic in satn_pdma_cb() is now no longer required since esp_do_dma() can be used as a direct replacement. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 75 +-- include/hw/scsi/esp.h | 1 - 2 files changed, 43 insertions(+), 33 de

[PATCH 44/88] esp.c: convert esp_reg_write() do_cmd path to check for SCSI phase instead

Currently do_cmd is used to determine whether MESSAGE OUT and COMMAND phase data is being accumulated in cmdfifo. Update esp_do_dma() to check directly for these two SCSI phases instead. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff

[PATCH 18/88] esp.c: don't clear RFLAGS register when DMA is complete

The internal state of the ESP sequencer is not affected when raising an interrupt to indicate the end of a DMA transfer. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 1 - 1 file changed, 1 deletion(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c index a4a1f41a40..5b9c3f1e5e 100644 --- a/h

[PATCH 46/88] esp.c: untangle MESSAGE OUT and COMMAND phase logic in esp_do_dma()

This makes it clearer that ATN is asserted until the end of the next TI command in the MESSAGE OUT phase. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 48 ++-- 1 file changed, 26 insertions(+), 22 deletions(-) diff --git a/hw/scsi/esp.c b/hw/sc

[PATCH 52/88] esp.c: move CMD_SELATNS end of command logic to esp_do_dma() and do_dma_pdma_cb()

The special logic in satn_stop_pdma_cb() is now no longer required since esp_do_dma() can be used as a direct replacement. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 82 +-- include/hw/scsi/esp.h | 1 - 2 files changed, 41 insertions(+),

[PATCH 71/88] esp.c: don't clear the SCSI phase when reading ESP_RINTR

According to the documentation ESP_RSTAT is cleared (except the STAT_TC bit) when ESP_RINTR is read. This should not include the SCSI bus phase bits which are currently live from the SCSI bus, otherwise the current SCSI phase is lost when clearing an end-of-transfer interrupt. Signed-off-by: Mark

[PATCH 69/88] esp.c: consolidate end of command sequence after ICCS command

The end of command sequences for the ICCS command are currently different between the DMA and non-DMA versions, and also different from the description in the datasheet. Update the sequence so that only INTR_FC is asserted in both cases, and keep all the logic in esp_do_dma() and esp_do_nodma() ra

[PATCH 03/88] esp.c: add FIFO wraparound support to esp_fifo_pop_buf()

The fifo8_pop_buf() function returns a pointer to the FIFO buffer up to the specified length. Since the FIFO buffer is modelled as an array then once the FIFO wraps around, only the continuous portion of the buffer can be returned. In future the use of continuous and unaligned accesses will advanc

[PATCH 4/4] esp-pci.c: set DMA_STAT_BCMBLT when BLAST command issued

Even though the BLAST command isn't fully implemented in QEMU, the DMA_STAT_BCMBLT bit should be set after the command has been issued to indicate that the command has completed. This fixes an issue with the DC390 DOS driver which issues the BLAST command as part of its normal error recovery rout

[PATCH 00/88] esp: rework ESP emulation to use a SCSI phase-based state machine

The ESP SCSI chip fundamentally consists of a FIFO for transferring data to/from the SCSI bus along with a command sequencer which automates various processes such as selection, message/command transfer and data transfer. What makes this chip particularly interesting is that the FIFO is also used

[PATCH 2/4] esp-pci.c: generate PCI interrupt from separate ESP and PCI sources

The am53c974/dc390 PCI interrupt has two separate sources: the first is from the internal ESP device, and the second is from the PCI DMA transfer logic. Update the ESP interrupt handler so that it sets DMA_STAT_SCSIINT rather than driving the PCI IRQ directly, and introduce a new esp_pci_update_ir

[PATCH 3/4] esp-pci.c: synchronise setting of DMA_STAT_DONE with ESP completion interrupt

The setting of DMA_STAT_DONE at the end of a DMA transfer can be configured to generate an interrupt, however the Linux driver manually checks for DMA_STAT_DONE being set and if it is, considers that a DMA transfer has completed. If DMA_STAT_DONE is set but the ESP device isn't indicating an inte

[PATCH 04/88] esp: remove FIFO clear from esp_select()

The FIFO contents should not be affected by performing SCSI target selection. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 1 - 1 file changed, 1 deletion(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c index 8d8f6a817a..89fce05e58 100644 --- a/hw/scsi/esp.c +++ b/hw/scsi/esp.c @@ -220,7 +

[PATCH 61/88] esp.c: remove DATA IN phase logic when reading from FIFO

Whilst the FIFO is used a storage buffer for both DMA and non-DMA requests, the loading and unloading is managed directly issuing commands to the ESP. As a result there is no need to manually invoke the non-DMA command handler. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 11 ---

[PATCH 48/88] esp.c: untangle MESSAGE OUT and COMMAND phase logic in esp_do_nodma()

This makes it clearer that ATN is asserted until the end of the next TI command in the MESSAGE OUT phase. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 40 +++- 1 file changed, 23 insertions(+), 17 deletions(-) diff --git a/hw/scsi/esp.c b/hw/scsi/esp.c

[PATCH 59/88] esp.c: separate logic based upon ESP command in esp_transfer_data()

The handling of the INTR_FC and INTR_BS bits is different depending upon the last command executed by the ESP. Note that currently INTR_FC is managed elsewhere, but that will change soon. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 32 ++-- 1 file changed, 26

[PATCH 63/88] esp.c: remove unneeded ti_cmd field

According to the datasheet the previous ESP command remains in the ESP_CMD register, which caused a problem when consecutive TI commands were issued as it becomes impossible for the state machine to know when the first TI command finishes. This was the original reason for introducing the ti_cmd fi

[PATCH 1/4] esp-pci.c: use correct address register for PCI DMA transfers

The current code in esp_pci_dma_memory_rw() sets the DMA address to the value of the DMA_SPA (Starting Physical Address) register which is incorrect: this means that for each callback from the SCSI layer the DMA address is set back to the starting address. In the case where only a single SCSI call

[PATCH 56/88] esp.c: remove unused PDMA callback implementation

Note that this is a migration break for the q800 machine because the extra PDMA information is no longer included. Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 55 --- include/hw/scsi/esp.h | 6 - 2 files changed, 5 insertions(+), 56 de

[PATCH 0/4] esp-pci: fixes for Linux and MS-DOS

This series contains fixes for the esp-pci device (am53c974 or dc390) for a few issues spotted whilst testing the previous ESP series. Patches 1-3 are fixes for issues found by Helge/Guenter whilst testing the hppa C3700 machine with the amd53c974/dc390 devices under Linux, whilst patch 4 fixes an

[PATCH 27/88] esp.c: update end of transfer logic at the end of esp_transfer_data()

Since esp_dma_done() is called in both cases, and ti_size cannot be zero (otherwise esp_command_complete() would have been called instead), replace the conditional logic with a single call to esp_dma_done(). Signed-off-by: Mark Cave-Ayland --- hw/scsi/esp.c | 14 -- 1 file changed, 4

Re: [PULL 13/13] tests/avocado: remove skips from replay_kernel

Thomas Huth writes: > On 08/01/2024 16.13, Alex Bennée wrote: >> With the latest fixes for #2010 and #2013 these tests look pretty >> stable now. Of course the only way to be really sure is to run it in >> the CI infrastructure and see what breaks. >> Acked-by: Pavel Dovgalyuk >> Signed-off-by:

Re: [PATCH 0/2] Export debug triggers as an extension

On Fri, 2024-01-12 at 13:52 +1000, Alistair Francis wrote: > On Thu, Jan 11, 2024 at 5:20 AM Daniel Henrique Barboza > wrote: > > > > Himanshu, > > > > We spoke offline but let's make everyone aware: > > > > - 'sdtrig' should be marked with 'x-' and be an experimental > > extension since > > th

test-x86-cpuid-compat qtest produces warnings on TCG

When running under TCG, the 'test-x86-cpuid-compat' qtest produces warnings. You can see these for instance in the x64-freebsd-13-build CI job; here's an example: https://gitlab.com/qemu-project/qemu/-/jobs/5918818568 24/837 qemu:qtest+qtest-x86_64 / qtest-x86_64/test-x86-cpuid-compat O

[PATCH] hw/pci-host/raven.c: Mark raven_io_ops as implementing unaligned accesses

The raven_io_ops MemoryRegionOps is the only one in the source tree which sets .valid.unaligned to indicate that it should support unaligned accesses and which does not also set .impl.unaligned to indicate that its read and write functions can do the unaligned handling themselves. This is a proble

Re: [PATCH] hw/arm/virt: Consolidate valid CPU types

On Thu, Jan 11 2024, Gavin Shan wrote: > It's found that some of the CPU type names in the array of valid > CPU types are invalid because their corresponding classes aren't > registered, as reported by Peter Maydell. > > [gshan@gshan build]$ ./qemu-system-arm -machine virt -cpu cortex-a9 > qemu-s

Re: [PATCH] hw/arm/virt: Consolidate valid CPU types

On Thu, 11 Jan 2024 at 05:11, Gavin Shan wrote: > > It's found that some of the CPU type names in the array of valid > CPU types are invalid because their corresponding classes aren't > registered, as reported by Peter Maydell. > > [gshan@gshan build]$ ./qemu-system-arm -machine virt -cpu cortex-a

Re: [PATCH v2 23/43] target/riscv: Remove misa_mxl validation

Alistair Francis writes: > On Tue, Jan 9, 2024 at 1:43 AM Alex Bennée wrote: >> >> Alex Bennée writes: >> >> > From: Akihiko Odaki >> > >> > It is initialized with a simple assignment and there is little room for >> > error. In fact, the validation is even more complex. >> > >> > Signed-off-by

[PATCH] ui/clipboard: avoid crash upon request when clipboard peer is not initialized

With VNC, it can be that a client sends a VNC_MSG_CLIENT_CUT_TEXT message before sending a VNC_MSG_CLIENT_SET_ENCODINGS message with VNC_ENCODING_CLIPBOARD_EXT for configuring the clipboard extension. This means that qemu_clipboard_request() can be reached (via vnc_client_cut_text_ext()) before vn

[PULL v2 00/22] testing and misc updates

The following changes since commit 3fab43d2f638b98765b0f8ef033ec93e3ea6b809: Merge tag 'firmware/edk2-20231213-pull-request' of https://gitlab.com/kraxel/qemu into staging (2024-01-11 15:19:42 +) are available in the Git repository at: https://gitlab.com/stsquad/qemu.git tags/pull-testi

[PATCH v5 3/8] target/riscv: move 'cbop_blocksize' to riscv_cpu_properties[]

Do the same we did with 'cbom_blocksize' in the previous patch. Signed-off-by: Daniel Henrique Barboza Reviewed-by: Alistair Francis Tested-by: Vladimir Isaev --- target/riscv/cpu.c | 38 +- 1 file changed, 37 insertions(+), 1 deletion(-) diff --git a/targe

[PATCH v5 2/8] target/riscv: move 'cbom_blocksize' to riscv_cpu_properties[]

After adding a KVM finalize() implementation, turn cbom_blocksize into a class property. Follow the same design we used with 'vlen' and 'elen'. The duplicated 'cbom_blocksize' KVM property can be removed from kvm_riscv_add_cpu_user_properties(). Signed-off-by: Daniel Henrique Barboza Reviewed-by

[PATCH v5 8/8] target/riscv/cpu.c: move 'marchid' to riscv_cpu_properties[]

Keep all class properties in riscv_cpu_properties[]. Signed-off-by: Daniel Henrique Barboza Reviewed-by: Alistair Francis Tested-by: Vladimir Isaev --- target/riscv/cpu.c | 110 +++-- 1 file changed, 57 insertions(+), 53 deletions(-) diff --git a/target

[PATCH v5 1/8] target/riscv: create finalize_features() for KVM

To turn cbom_blocksize and cboz_blocksize into class properties we need KVM specific changes. KVM is creating its own version of these options with a customized setter() that prevents users from picking an invalid value during init() time. This comes at the cost of duplicating each option that KVM

[PATCH v5 7/8] target/riscv/cpu.c: move 'mimpid' to riscv_cpu_properties[]

Keep all class properties in riscv_cpu_properties[]. Signed-off-by: Daniel Henrique Barboza Reviewed-by: Alistair Francis Tested-by: Vladimir Isaev --- target/riscv/cpu.c | 68 -- 1 file changed, 36 insertions(+), 32 deletions(-) diff --git a/target

[PATCH v5 5/8] target/riscv: remove riscv_cpu_options[]

The array is empty and can be removed. Signed-off-by: Daniel Henrique Barboza Reviewed-by: Alistair Francis Tested-by: Vladimir Isaev --- target/riscv/cpu.c | 5 - target/riscv/cpu.h | 1 - target/riscv/kvm/kvm-cpu.c | 9 - target/riscv/tcg/tcg-cpu.c | 4 4 fil

[PATCH v5 4/8] target/riscv: move 'cboz_blocksize' to riscv_cpu_properties[]

And remove the now unused kvm_cpu_set_cbomz_blksize() setter. Signed-off-by: Daniel Henrique Barboza Reviewed-by: Alistair Francis Tested-by: Vladimir Isaev --- target/riscv/cpu.c | 38 +- target/riscv/kvm/kvm-cpu.c | 28

[PATCH v5 0/8] target/riscv: deprecate riscv_cpu_options[]

Hi, This version is a resend of patches 10 to 17 from v4, reviewed-by and tested-by tags added, rebased with Alistair's riscv-to-apply.next. Patches 01 to 09 of v4 are already queued. All patches acked. Changes from v4: - patches 1 to 9: already queued - patches 10 to 17: reviewed-by and tested-

[PATCH v5 6/8] target/riscv/cpu.c: move 'mvendorid' to riscv_cpu_properties[]

Keep all class properties in riscv_cpu_properties[]. Signed-off-by: Daniel Henrique Barboza Reviewed-by: Alistair Francis Tested-by: Vladimir Isaev --- target/riscv/cpu.c | 69 +- 1 file changed, 37 insertions(+), 32 deletions(-) diff --git a/target

Re: [PATCH v2 2/2] hmp: Drop unknown feature and status bits

Hyman Huang writes: > The QMP command "x-query-virtio-status" outputs the full > feature and status bit information, so there is no need > to maintain it in the HMP output; drop it. > > Signed-off-by: Hyman Huang I'm not sure hiding unknown bits from HMP users is a good idea. Sure, you can use

Re: [PATCH v2 1/2] qapi/virtio: Keep feature and status bits in the QMP output

Hyman Huang writes: > Maintain the feature and status bits in the x-query-virtio-status > output and, as usual, add human-readable output only in HMP. > > Applications may find it useful to compare features and status > information directly. An upper application, for example, could > use the QMP

Re: [PATCH] ui/clipboard: avoid crash upon request when clipboard peer is not initialized

Am 12.01.24 um 14:55 schrieb Fiona Ebner: > > Then when noVNC sends a VNC_MSG_CLIENT_CUT_TEXT message (it did set > the feature correctly, so the check added by your patch passes), Sorry, forgot to adapt this part. This should read "it did set the VNC_FEATURE_CLIPBOARD_EXT feature correctly, so a

Re: Goldfish TTY enhancement

Hi Jason, Le 11/01/2024 à 19:58, Jason Thorpe a écrit : On Jan 10, 2024, at 8:01 AM, Philippe Mathieu-Daudé wrote: IIUC Goldfish virtual HW is maintained externally by Google https://android.googlesource.com/platform/external/qemu/+/master/docs/GOLDFISH-VIRTUAL-HARDWARE.TXT I suppose the sp

Re: [PULL 0/6] Firmware/edk2 20231213 patches

11.01.2024 19:28, Gerd Hoffmann: .. Yes, Cc'ing qemu-stable was the intention, thanks for fixing it up. I'd leave it to the stable maintainer(s). If they prefer a respin with Cc qemu-stable added to all patches I surely can do that. If being notified with this reply is good enough I'm happy to

Re: [PATCH 01/12] plugins: implement inline operation with cpu_index offset

On 1/12/24 02:04, Richard Henderson wrote: On 1/12/24 01:23, Pierrick Bouvier wrote: Instead of working on a fixed memory location, allow to index it based on cpu_index and a given offset (ptr + cpu_index * offset). Current semantic is not modified as we use a 0 offset, thus inline operation sti

Re: [PULL 0/6] Firmware/edk2 20231213 patches

12.01.2024 17:27, Michael Tokarev: There's no requirement to have Cc: qemu-stable tags on the patches. This tagging is only to easily find changes which are supposed to be picked up for stable, nothing more, so it's completely optional. An additional note: without this Cc by Peter, I would've n

Re: [PATCH v10 0/9] Introduce model for IBM's FSI

Hello Ninad, This is looking much better. I have a couple more comments. Please wait a little before sending a respin ! :) and please run make check and script/checkpatch.pl. On 1/11/24 00:15, Ninad Palsule wrote: Hello, Please review the patch-set version 10. I have incorporated review comme

Re: [PATCH v10 1/9] hw/fsi: Introduce IBM's Local bus and scratchpad

On 1/11/24 00:15, Ninad Palsule wrote: This is a part of patchset where IBM's Flexible Service Interface is introduced. The LBUS is modelled to maintain mapped memory for the devices. The memory is mapped after CFAM config, peek table and FSI slave registers. The scratchpad provides a set of no

Re: [PATCH v2 2/2] hmp: Drop unknown feature and status bits

On Fri, Jan 12, 2024 at 10:04 PM Markus Armbruster wrote: > Hyman Huang writes: > > > The QMP command "x-query-virtio-status" outputs the full > > feature and status bit information, so there is no need > > to maintain it in the HMP output; drop it. > > > > Signed-off-by: Hyman Huang > > I'm no

Re: [PATCH v10 3/9] hw/fsi: Introduce IBM's cfam

On 1/11/24 00:15, Ninad Palsule wrote: This is a part of patchset where IBM's Flexible Service Interface is introduced. The Common FRU Access Macro (CFAM), an address space containing various "engines" that drive accesses on busses internal and external to the POWER chip. Examples include the SB

Re: [PATCH v10 2/9] hw/fsi: Introduce IBM's FSI Bus and FSI slave

On 1/11/24 00:15, Ninad Palsule wrote: This is a part of patchset where FSI bus is introduced. The FSI bus is a simple bus where FSI master is attached. The FSI slave: The slave is the terminal point of the FSI bus for FSI symbols addressed to it. Slaves can be cascaded off of one another. The

Re: [PATCH v2 1/2] qapi/virtio: Keep feature and status bits in the QMP output

On Fri, Jan 12, 2024 at 10:10 PM Markus Armbruster wrote: > Hyman Huang writes: > > > Maintain the feature and status bits in the x-query-virtio-status > > output and, as usual, add human-readable output only in HMP. > > > > Applications may find it useful to compare features and status > > info

[PATCH V2 06/11] migration: MigrationNotifyFunc

Define MigrationNotifyFunc to improve type safety and simplify migration notifiers. Signed-off-by: Steve Sistare --- hw/net/virtio-net.c | 4 +--- hw/vfio/migration.c | 3 +-- include/migration/misc.h | 5 - migration/migration.c| 4 ++-- net/vhost-vdpa.c | 6 ++ ui

[PATCH V2 01/11] notify: pass error to notifier with return

Pass an error object as the third parameter to "notifier with return" notifiers, so clients no longer need to bundle an error object in the opaque data. The new parameter is used in a later patch. Signed-off-by: Steve Sistare --- hw/virtio/vhost-user.c | 2 +- hw/virtio/virtio-balloon.c | 3

[PATCH V2 04/11] migration: remove migration_in_postcopy parameter

Delete the explicit MigrationState parameter from migration_in_postcopy, so we can eliminate MigrationState from notifiers in a later patch. No functional change. Signed-off-by: Steve Sistare --- include/migration/misc.h | 2 +- migration/migration.c| 27 --- ui/spic

[PATCH V2 05/11] migration: MigrationEvent for notifiers

Passing MigrationState to notifiers is unsound because they could access unstable migration state internals or even modify the state. Instead, pass the minimal info needed in a new MigrationEvent struct, which could be extended in the future if needed. Suggested-by: Peter Xu Signed-off-by: Steve

[PATCH V2 08/11] migration: refactor migrate_fd_connect failures

Move common code for the error path in migrate_fd_connect to a shared fail label. No functional change. Signed-off-by: Steve Sistare --- migration/migration.c | 22 +++--- 1 file changed, 11 insertions(+), 11 deletions(-) diff --git a/migration/migration.c b/migration/migration

<    1   2   3   4   >