[PATCH v3 22/24] accel: build qtest modular

Allow building accelerators as module. Start with qtest as first user. Signed-off-by: Gerd Hoffmann --- accel/qtest/meson.build | 8 ++-- meson.build | 6 ++ 2 files changed, 8 insertions(+), 6 deletions(-) diff --git a/accel/qtest/meson.build b/accel/qtest/meson.build index

[PATCH v3 21/24] accel: add qtest module annotations

Add module annotations for qtest so autoloading works. Signed-off-by: Gerd Hoffmann --- accel/qtest/qtest.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/accel/qtest/qtest.c b/accel/qtest/qtest.c index edb29f6fa4c0..7e6b8110d52b 100644 --- a/accel/qtest/qtest.c +++ b/accel/qtest/qtest.c

[PATCH v3 18/24] modules: check arch on qom lookup

With target-specific modules we can have multiple modules implementing the same object. Therefore we have to check the target arch on lookup to find the correct module. Signed-off-by: Gerd Hoffmann --- util/module.c | 6 ++ 1 file changed, 6 insertions(+) diff --git a/util/module.c b/util/

[PATCH v3 15/24] modules: use modinfo for qemu opts load

Use module database to figure which module adds given QemuOpts group. Signed-off-by: Gerd Hoffmann --- softmmu/vl.c| 17 - stubs/module-opts.c | 4 util/module.c | 19 +++ 3 files changed, 19 insertions(+), 21 deletions(-) diff --git a/softmmu

[PATCH v3 14/24] modules: use modinfo for qom load

Use module database to figure which module implements a given QOM type. Drop hard-coded object list. Signed-off-by: Gerd Hoffmann --- util/module.c | 77 --- 1 file changed, 24 insertions(+), 53 deletions(-) diff --git a/util/module.c b/util/modul

[PATCH v3 11/24] modules: add s390x module annotations

Signed-off-by: Gerd Hoffmann --- hw/s390x/virtio-ccw-gpu.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/hw/s390x/virtio-ccw-gpu.c b/hw/s390x/virtio-ccw-gpu.c index 75a9e4bb3908..5868a2a07093 100644 --- a/hw/s390x/virtio-ccw-gpu.c +++ b/hw/s390x/virtio-ccw-gpu.c @@ -59,6 +59,7 @@ static

[PATCH v3 12/24] modules: add block module annotations

Signed-off-by: Gerd Hoffmann --- block/iscsi-opts.c | 1 + 1 file changed, 1 insertion(+) diff --git a/block/iscsi-opts.c b/block/iscsi-opts.c index afaf8837d6c1..4f2da405e645 100644 --- a/block/iscsi-opts.c +++ b/block/iscsi-opts.c @@ -68,3 +68,4 @@ static void iscsi_block_opts_init(void) }

[PATCH v3 24/24] accel: build tcg modular

Build tcg accel ops as module. Which is only a small fraction of tcg. Also only x86 for now. Signed-off-by: Gerd Hoffmann --- accel/tcg/meson.build | 5 - meson.build | 14 +- 2 files changed, 17 insertions(+), 2 deletions(-) diff --git a/accel/tcg/meson.build b/accel

[PATCH v3 17/24] modules: check arch and block load on mismatch

Add module_allow_arch() to set the target architecture. In case a module is limited to some arch verify arches match and ignore the module if not. Signed-off-by: Gerd Hoffmann --- include/qemu/module.h | 1 + softmmu/vl.c | 3 +++ util/module.c | 29 +++

[PATCH v3 09/24] modules: add ccid module annotations

Signed-off-by: Gerd Hoffmann --- hw/usb/ccid-card-emulated.c | 1 + hw/usb/ccid-card-passthru.c | 1 + 2 files changed, 2 insertions(+) diff --git a/hw/usb/ccid-card-emulated.c b/hw/usb/ccid-card-emulated.c index 5c76bed77aa0..6c8c0355e099 100644 --- a/hw/usb/ccid-card-emulated.c +++ b/hw/usb/cc

[PATCH v3 04/24] modules: add qxl module annotations

Signed-off-by: Gerd Hoffmann --- hw/display/qxl.c | 4 1 file changed, 4 insertions(+) diff --git a/hw/display/qxl.c b/hw/display/qxl.c index 6e1f8ff1b2a7..84f99088e0a0 100644 --- a/hw/display/qxl.c +++ b/hw/display/qxl.c @@ -2522,6 +2522,7 @@ static const TypeInfo qxl_primary_info = {

[PATCH v3 07/24] modules: add audio module annotations

Signed-off-by: Gerd Hoffmann --- audio/spiceaudio.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/audio/spiceaudio.c b/audio/spiceaudio.c index 999bfbde47c5..a8d370fe6f31 100644 --- a/audio/spiceaudio.c +++ b/audio/spiceaudio.c @@ -317,3 +317,5 @@ static void register_audio_spice(void)

[PATCH v3 23/24] accel: add tcg module annotations

Add module annotations for tcg so autoloading works. Signed-off-by: Gerd Hoffmann --- accel/tcg/tcg-accel-ops.c | 1 + accel/tcg/tcg-all.c | 1 + 2 files changed, 2 insertions(+) diff --git a/accel/tcg/tcg-accel-ops.c b/accel/tcg/tcg-accel-ops.c index 7191315aeed4..1a8e8390bd60 100644 ---

[PATCH v3 19/24] modules: target-specific module build infrastructure

Signed-off-by: Gerd Hoffmann --- meson.build | 36 1 file changed, 36 insertions(+) diff --git a/meson.build b/meson.build index 9cf50a50d39a..e822477a231a 100644 --- a/meson.build +++ b/meson.build @@ -1781,6 +1781,7 @@ user_ss = ss.source_set() util_ss = s

[PATCH v3 16/24] modules: add tracepoints

One for module load and one for qom type lookup. Signed-off-by: Gerd Hoffmann --- util/module.c | 3 +++ util/trace-events | 4 2 files changed, 7 insertions(+) diff --git a/util/module.c b/util/module.c index a9ec2da9972e..acaaecad56c9 100644 --- a/util/module.c +++ b/util/module.c @@

[PATCH v3 13/24] modules: use modinfo for dependencies

Use module database for module dependencies. Drop hard-coded dependency list. Signed-off-by: Gerd Hoffmann --- util/module.c | 55 --- 1 file changed, 21 insertions(+), 34 deletions(-) diff --git a/util/module.c b/util/module.c index 8d3e8275b9f7.

[PATCH v3 08/24] modules: add usb-redir module annotations

Signed-off-by: Gerd Hoffmann --- hw/usb/redirect.c | 1 + 1 file changed, 1 insertion(+) diff --git a/hw/usb/redirect.c b/hw/usb/redirect.c index 6a75b0dc4ab2..4ec9326e0582 100644 --- a/hw/usb/redirect.c +++ b/hw/usb/redirect.c @@ -2608,6 +2608,7 @@ static const TypeInfo usbredir_dev_info = {

[PATCH v3 20/24] accel: autoload modules

Call module_object_class_by_name() instead of object_class_by_name() for objects possibly implemented as module Signed-off-by: Gerd Hoffmann --- accel/accel-common.c | 2 +- accel/accel-softmmu.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/accel/accel-common.c b/accel

[PATCH v3 00/24] modules: add meta-data database

This patch series adds support for module meta-data. Today this is either hard-coded in qemu (see qemu_load_module_for_opts) or handled with manually maintained lists in util/module (see module_deps[] and qom_modules[]). This series replaced that scheme with annotation macros, so the meta-data ca

[PATCH v3 10/24] modules: add ui module annotations

Signed-off-by: Gerd Hoffmann --- ui/egl-headless.c | 4 ui/gtk.c | 4 ui/sdl2.c | 4 ui/spice-app.c| 3 +++ ui/spice-core.c | 5 + 5 files changed, 20 insertions(+) diff --git a/ui/egl-headless.c b/ui/egl-headless.c index da377a74af69..75404e0e8700 10064

[PATCH v3 06/24] modules: add chardev module annotations

Signed-off-by: Gerd Hoffmann --- chardev/baum.c | 1 + chardev/spice.c | 4 2 files changed, 5 insertions(+) diff --git a/chardev/baum.c b/chardev/baum.c index 5deca778bc44..79d618e35045 100644 --- a/chardev/baum.c +++ b/chardev/baum.c @@ -680,6 +680,7 @@ static const TypeInfo char_braille

[PATCH v3 05/24] modules: add virtio-gpu module annotations

Signed-off-by: Gerd Hoffmann --- hw/display/vhost-user-gpu-pci.c | 1 + hw/display/vhost-user-gpu.c | 1 + hw/display/vhost-user-vga.c | 1 + hw/display/virtio-gpu-base.c| 1 + hw/display/virtio-gpu-gl.c | 3 +++ hw/display/virtio-gpu-pci-gl.c | 3 +++ hw/display/virtio-gpu-pci.

[PATCH v3 02/24] modules: collect module meta-data

Add script to collect the module meta-data from the source code, store the results in *.modinfo files. Signed-off-by: Gerd Hoffmann --- scripts/modinfo-collect.py | 67 ++ meson.build| 11 +++ 2 files changed, 78 insertions(+) create mode

[PATCH v3 03/24] modules: generate modinfo.c

Add script to generate C source with a small database containing the module meta-data. Signed-off-by: Gerd Hoffmann --- scripts/modinfo-generate.py | 84 + include/qemu/module.h | 17 softmmu/vl.c| 4 ++ util/module.c

[PATCH v3 01/24] modules: add modinfo macros

Add macros for module info annotations. Instead of having that module meta-data stored in lists in util/module.c place directly in the module source code. Signed-off-by: Gerd Hoffmann --- include/qemu/module.h | 25 + 1 file changed, 25 insertions(+) diff --git a/includ

[PATCH RFC] meson: add option to use zstd for qcow2 compression by default

Recently we added support of zstd to qcow2 format, as zstd seems to be better than zlib in general, and which is important (as qcow2 compression used mostly for backups) compressed writes are faster with zstd. Let's add a build option to use zstd by default. Signed-off-by: Vladimir Sementsov-Ogie

[PATCH v2 10/11] hw/nvme: reimplement format nvm to allow cancellation

From: Klaus Jensen Prior to this patch, the aios associated with broadcast format are submitted anonymously (no aiocb reference saved from the blk_aio call). Fix this by formatting the namespaces one after another, saving a reference to the aiocb for each. Signed-off-by: Klaus Jensen --- hw/n

[PATCH v2 08/11] hw/nvme: reimplement the copy command to allow aio cancellation

From: Klaus Jensen Before this patch the code would issue several aios simultaneously without saving a reference to the aiocb. Without the aiocb reference the individual copies cannot be canceled. Fix this by issuing copies of the ranges one after another. Signed-off-by: Klaus Jensen --- hw/n

Re: [PATCH] hw/nvme: fix missing check for PMR capability

On Jun 7 11:47, Klaus Jensen wrote: From: Klaus Jensen Qiang Liu reported that an access on an unknown address is triggered in memory_region_set_enabled because a check on CAP.PMRS is missing for the PMRCTL register write when no PMR is configured. Cc: qemu-sta...@nongnu.org Fixes: 75c3c9de96

[PATCH v2 06/11] hw/nvme: use prinfo directly in nvme_check_prinfo and nvme_dif_check

From: Klaus Jensen The nvme_check_prinfo() and nvme_dif_check() functions operate on the 16 bit "control" member of the NvmeCmd. These functions do not otherwise operate on an NvmeCmd or an NvmeRequest, so change them to expect the actual 4 bit PRINFO field and add constants that work on this fie

[PATCH v2 04/11] hw/nvme: save reftag when generating pi

From: Klaus Jensen Prepare nvme_dif_pract_generate_dif() and nvme_dif_check() to be callable in smaller increments by making the reftag a pointer parameter updated by the function. Signed-off-by: Klaus Jensen --- hw/nvme/nvme.h | 4 ++-- hw/nvme/ctrl.c | 10 +- hw/nvme/dif.c | 22 +++

[PATCH v2 11/11] Partially revert "hw/block/nvme: drain namespaces on sq deletion"

From: Klaus Jensen This partially reverts commit 98f84f5a4eca5c03e32fff20f246d9b4b96d6422. Since all "multi aio" commands are now reimplemented to properly track the nested aiocbs, we can revert the "hack" that was introduced to make sure all requests we're properly drained upon sq deletion. Th

[PATCH v2 07/11] hw/nvme: add dw0/1 to the req completion trace event

From: Klaus Jensen Some commands report additional useful information in dw0 and dw1 of the completion queue entry. Add them to the trace. Signed-off-by: Klaus Jensen --- hw/nvme/ctrl.c | 2 ++ hw/nvme/trace-events | 2 +- 2 files changed, 3 insertions(+), 1 deletion(-) diff --git a/hw

[PATCH v2 09/11] hw/nvme: reimplement zone reset to allow cancellation

From: Klaus Jensen Prior to this patch, the aios associated with zone reset are submitted anonymously (no reference saved to the aiocb from the blk_aio call). Fix this by resetting the zones one after another, saving a reference to the aiocb for each reset. Signed-off-by: Klaus Jensen --- hw/

[PATCH v2 05/11] hw/nvme: remove assert from nvme_get_zone_by_slba

From: Klaus Jensen Make nvme_get_zone_by_slba() return NULL if the slba is out of range. This allows the function to be used without guarding the call with a call to nvme_check_bounds(), in preparation for the next patch. Add asserts after calling nvme_get_zone_by_slba() instead. Signed-off-by:

[PATCH v2 02/11] hw/nvme: add nvme_block_status_all helper

From: Klaus Jensen Pull the gist of nvme_check_dulbe() into a helper function. This is in preparation for dsm refactoring. Signed-off-by: Klaus Jensen --- hw/nvme/ctrl.c | 41 - 1 file changed, 28 insertions(+), 13 deletions(-) diff --git a/hw/nvme/ctrl

[PATCH v2 03/11] hw/nvme: reimplement dsm to allow cancellation

From: Klaus Jensen Prior to this patch, a loop was used to issue multiple "fire and forget" aios for each range in the command. Without a reference to the aiocb returned from the blk_aio_pdiscard calls, the aios cannot be canceled. Fix this by processing the ranges one after another. As a bonus

[PATCH v2 01/11] hw/nvme: reimplement flush to allow cancellation

From: Klaus Jensen Prior to this patch, a broadcast flush would result in submitting multiple "fire and forget" aios (no reference saved to the aiocbs returned from the blk_aio_flush calls). Fix this by issuing the flushes one after another. Signed-off-by: Klaus Jensen --- hw/nvme/nvme.h

[PATCH v2 00/11] hw/nvme: reimplement all multi-aio commands with custom aiocbs

From: Klaus Jensen This series reimplements flush, dsm, copy, zone reset and format nvm to allow cancellation. I posted an RFC back in March ("hw/block/nvme: convert ad-hoc aio tracking to aiocb") and I've applied some feedback from Stefan and reimplemented the remaining commands. The basic idea

[PATCH v2] hw/nvme: fix pin-based interrupt behavior (again)

From: Klaus Jensen Jakub noticed[1] that, when using pin-based interrupts, the device will unconditionally deasssert when any CQEs are acknowledged. However, the pin should not be deasserted if other completion queues still holds unacknowledged CQEs. The bug is an artifact of commit ca247d35098d

Re: [PATCH] hw/nvme: fix pin-based interrupt behavior (again)

On Jun 17 07:50, Keith Busch wrote: On Thu, Jun 17, 2021 at 12:08:20PM +0200, Klaus Jensen wrote: if (cq->tail != cq->head) { +if (!pending) { +n->cq_pending++; +} You should check cq->irq_enabled before incrementing cq_pending. You don't want to leave the irq

Re: [PULL 28/34] nbd/client-connection: return only one io channel

15.06.2021 23:47, Eric Blake wrote: From: Vladimir Sementsov-Ogievskiy block/nbd doesn't need underlying sioc channel anymore. So, we can update nbd/client-connection interface to return only one top-most io channel, which is more straight forward. Signed-off-by: Vladimir Sementsov-Ogievskiy

Re: [PATCH] block/nvme: Fix VFIO_MAP_DMA failed: No space left on device

On 6/17/21 2:40 PM, Maxim Levitsky wrote: > On Mon, 2021-06-14 at 18:03 +0200, Philippe Mathieu-Daudé wrote: >> On 6/11/21 1:46 PM, Philippe Mathieu-Daudé wrote: >>> When the NVMe block driver was introduced (see commit bdd6a90a9e5, >>> January 2018), Linux VFIO_IOMMU_MAP_DMA ioctl was only returni

[PATCH] block/rbd: Add support for rbd image encryption

Starting from ceph Pacific, RBD has built-in support for image-level encryption. Currently supported formats are LUKS version 1 and 2. There are 2 new relevant librbd APIs for controlling encryption, both expect an open image context: rbd_encryption_format: formats an image (i.e. writes the LUKS

[PATCH 6/6] block/iscsi: Do not force-cap *pnum

bdrv_co_block_status() does it for us, we do not need to do it here. The advantage of not capping *pnum is that bdrv_co_block_status() can cache larger data regions than requested by its caller. Signed-off-by: Max Reitz --- block/iscsi.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/blo

[PATCH 4/6] block/gluster: Do not force-cap *pnum

bdrv_co_block_status() does it for us, we do not need to do it here. The advantage of not capping *pnum is that bdrv_co_block_status() can cache larger data regions than requested by its caller. Signed-off-by: Max Reitz --- block/gluster.c | 7 --- 1 file changed, 4 insertions(+), 3 deletio

[PATCH 2/6] block: block-status cache for data regions

As we have attempted before (https://lists.gnu.org/archive/html/qemu-devel/2019-01/msg06451.html, "file-posix: Cache lseek result for data regions"; https://lists.nongnu.org/archive/html/qemu-block/2021-02/msg00934.html, "file-posix: Cache next hole"), this patch seeks to reduce the number of SEEK_

[PATCH 5/6] block/nbd: Do not force-cap *pnum

bdrv_co_block_status() does it for us, we do not need to do it here. The advantage of not capping *pnum is that bdrv_co_block_status() can cache larger data regions than requested by its caller. Signed-off-by: Max Reitz --- block/nbd.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff

[PATCH 3/6] block/file-posix: Do not force-cap *pnum

bdrv_co_block_status() does it for us, we do not need to do it here. The advantage of not capping *pnum is that bdrv_co_block_status() can cache larger data regions than requested by its caller. Signed-off-by: Max Reitz --- block/file-posix.c | 7 --- 1 file changed, 4 insertions(+), 3 dele

[PATCH 0/6] block: block-status cache for data regions

Hi, We’ve already had two attempts at introducing a block-status cache for data regions (https://lists.gnu.org/archive/html/qemu-devel/2019-01/msg06451.html, "file-posix: Cache lseek result for data regions"; https://lists.nongnu.org/archive/html/qemu-block/2021-02/msg00934.html, "file-posix: Cach

[PATCH 1/6] block: Drop BDS comment regarding bdrv_append()

There is a comment above the BDS definition stating care must be taken to consider handling newly added fields in bdrv_append(). Actually, this comment should have said "bdrv_swap()" as of 4ddc07cac (nine years ago), and in any case, bdrv_swap() was dropped in 8e419aefa (six years ago). So no suc

Re: [PATCH] hw/nvme: fix pin-based interrupt behavior (again)

On Thu, Jun 17, 2021 at 12:08:20PM +0200, Klaus Jensen wrote: > if (cq->tail != cq->head) { > +if (!pending) { > +n->cq_pending++; > +} You should check cq->irq_enabled before incrementing cq_pending. You don't want to leave the irq asserted when polled queues have

Re: [PATCH V3 4/6] block/rbd: migrate from aio to coroutines

On Wed, May 19, 2021 at 4:27 PM Peter Lieven wrote: > > Signed-off-by: Peter Lieven > --- > block/rbd.c | 255 ++-- > 1 file changed, 87 insertions(+), 168 deletions(-) > > diff --git a/block/rbd.c b/block/rbd.c > index 97a2ae4c84..0d8612a988 10064

Re: [PATCH] hw/nvme: fix pin-based interrupt behavior (again)

On 6/17/21 12:09 PM, Klaus Jensen wrote: On Jun 17 12:08, Klaus Jensen wrote: From: Klaus Jensen Jakub noticed[1] that, when using pin-based interrupts, the device will unconditionally deasssert when any CQEs are acknowledged. However, the pin should not be deasserted if other completion queue

Re: [PATCH] block/nvme: Fix VFIO_MAP_DMA failed: No space left on device

On Mon, 2021-06-14 at 18:03 +0200, Philippe Mathieu-Daudé wrote: > On 6/11/21 1:46 PM, Philippe Mathieu-Daudé wrote: > > When the NVMe block driver was introduced (see commit bdd6a90a9e5, > > January 2018), Linux VFIO_IOMMU_MAP_DMA ioctl was only returning > > -ENOMEM in case of error. The driver w

Re: [PATCH] hw/nvme: fix pin-based interrupt behavior (again)

On Jun 17 12:08, Klaus Jensen wrote: From: Klaus Jensen Jakub noticed[1] that, when using pin-based interrupts, the device will unconditionally deasssert when any CQEs are acknowledged. However, the pin should not be deasserted if other completion queues still holds unacknowledged CQEs. The bu

[PATCH] hw/nvme: fix pin-based interrupt behavior (again)

From: Klaus Jensen Jakub noticed[1] that, when using pin-based interrupts, the device will unconditionally deasssert when any CQEs are acknowledged. However, the pin should not be deasserted if other completion queues still holds unacknowledged CQEs. The bug is an artifact of commit ca247d35098d

Re: [PATCH v3 1/2] hw/nvme: fix endianess conversion and add controller list

On Jun 14 21:59, Gollu Appalanaidu wrote: Add the controller identifiers list CNS 0x13, available list of ctrls in NVM Subsystem that may or may not be attached to namespaces. In Identify Ctrl List of the CNS 0x12 and 0x13 no endian conversion for the nsid field. These two CNS values shows affe