From: "Dr. David Alan Gilbert" <dgilbert@redhat.com>

Define a new capability type 'VIRTIO_PCI_CAP_SHARED_MEMORY_CFG'
and the data structure 'virtio_pci_shm_cap' to go with it.
They allow defining shared memory regions with sizes and offsets
of 2^32 and more.
Multiple instances of the capability are allowed and distinguished
by a device-specific 'id'.

v2: Remove virtio_pci_shm_cap as virtio_pci_cap64 is used instead.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
(cherry picked from commit a5d628a3a3c5e60b98b15ffff197c36a77056115)
Signed-off-by: Antonio Caggiano <antonio.caggiano@collabora.com>
---
 hw/virtio/virtio-pci.c | 19 +++++++++++++++++++
 hw/virtio/virtio-pci.h |  4 ++++
 2 files changed, 23 insertions(+)

diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
index 750aa47ec1..8152d3c1b3 100644
--- a/hw/virtio/virtio-pci.c
+++ b/hw/virtio/virtio-pci.c
@@ -1162,6 +1162,25 @@ static int virtio_pci_add_mem_cap(VirtIOPCIProxy *proxy,
     return offset;
 }
 
+int virtio_pci_add_shm_cap(VirtIOPCIProxy *proxy,
+                           uint8_t bar, uint64_t offset, uint64_t length,
+                           uint8_t id)
+{
+    struct virtio_pci_cap64 cap = {
+        .cap.cap_len = sizeof cap,
+        .cap.cfg_type = VIRTIO_PCI_CAP_SHARED_MEMORY_CFG,
+    };
+    uint32_t mask32 = ~0;
+
+    cap.cap.bar = bar;
+    cap.cap.length = cpu_to_le32(length & mask32);
+    cap.length_hi = cpu_to_le32((length >> 32) & mask32);
+    cap.cap.offset = cpu_to_le32(offset & mask32);
+    cap.offset_hi = cpu_to_le32((offset >> 32) & mask32);
+    cap.cap.id = id;
+    return virtio_pci_add_mem_cap(proxy, &cap.cap);
+}
+
 static uint64_t virtio_pci_common_read(void *opaque, hwaddr addr,
                                        unsigned size)
 {
diff --git a/hw/virtio/virtio-pci.h b/hw/virtio/virtio-pci.h
index 2446dcd9ae..5e5c4a4c6d 100644
--- a/hw/virtio/virtio-pci.h
+++ b/hw/virtio/virtio-pci.h
@@ -252,4 +252,8 @@ void virtio_pci_types_register(const VirtioPCIDeviceTypeInfo *t);
  */
 unsigned virtio_pci_optimal_num_queues(unsigned fixed_queues);
 
+int virtio_pci_add_shm_cap(VirtIOPCIProxy *proxy,
+                           uint8_t bar, uint64_t offset, uint64_t length,
+                           uint8_t id);
+
 #endif
-- 
2.32.0

On Wed, Nov 10, 2021 at 05:42:20PM +0100, Antonio Caggiano wrote:
> From: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
> 
> Define a new capability type 'VIRTIO_PCI_CAP_SHARED_MEMORY_CFG'
> and the data structure 'virtio_pci_shm_cap' to go with it.
> They allow defining shared memory regions with sizes and offsets
> of 2^32 and more.
> Multiple instances of the capability are allowed and distinguished
> by a device-specific 'id'.
> 
> v2: Remove virtio_pci_shm_cap as virtio_pci_cap64 is used instead.
> 
> Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
> (cherry picked from commit a5d628a3a3c5e60b98b15ffff197c36a77056115)

Where's that commit? I think we should drop this, right?

> Signed-off-by: Antonio Caggiano <antonio.caggiano@collabora.com>
> ---
>  hw/virtio/virtio-pci.c | 19 +++++++++++++++++++
>  hw/virtio/virtio-pci.h |  4 ++++
>  2 files changed, 23 insertions(+)
> 
> diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
> index 750aa47ec1..8152d3c1b3 100644
> --- a/hw/virtio/virtio-pci.c
> +++ b/hw/virtio/virtio-pci.c
> @@ -1162,6 +1162,25 @@ static int virtio_pci_add_mem_cap(VirtIOPCIProxy *proxy,
>      return offset;
>  }
>  
> +int virtio_pci_add_shm_cap(VirtIOPCIProxy *proxy,
> +                           uint8_t bar, uint64_t offset, uint64_t length,
> +                           uint8_t id)
> +{
> +    struct virtio_pci_cap64 cap = {
> +        .cap.cap_len = sizeof cap,
> +        .cap.cfg_type = VIRTIO_PCI_CAP_SHARED_MEMORY_CFG,
> +    };
> +    uint32_t mask32 = ~0;
> +
> +    cap.cap.bar = bar;
> +    cap.cap.length = cpu_to_le32(length & mask32);
> +    cap.length_hi = cpu_to_le32((length >> 32) & mask32);
> +    cap.cap.offset = cpu_to_le32(offset & mask32);
> +    cap.offset_hi = cpu_to_le32((offset >> 32) & mask32);
> +    cap.cap.id = id;
> +    return virtio_pci_add_mem_cap(proxy, &cap.cap);


You don't need & mask32 I think. cpu_to_le32 will truncate
the value.


> +}
> +
>  static uint64_t virtio_pci_common_read(void *opaque, hwaddr addr,
>                                         unsigned size)
>  {
> diff --git a/hw/virtio/virtio-pci.h b/hw/virtio/virtio-pci.h
> index 2446dcd9ae..5e5c4a4c6d 100644
> --- a/hw/virtio/virtio-pci.h
> +++ b/hw/virtio/virtio-pci.h
> @@ -252,4 +252,8 @@ void virtio_pci_types_register(const VirtioPCIDeviceTypeInfo *t);
>   */
>  unsigned virtio_pci_optimal_num_queues(unsigned fixed_queues);
>  
> +int virtio_pci_add_shm_cap(VirtIOPCIProxy *proxy,
> +                           uint8_t bar, uint64_t offset, uint64_t length,
> +                           uint8_t id);
> +
>  #endif

So it's a new API, but where's the user?
I guess just include this patch with where-ever it's actually used.

> -- 
> 2.32.0

> Where's that commit? I think we should drop this, right? 

Yes, I will submit another version without that line.

> You don't need & mask32 I think. cpu_to_le32 will truncate
> the value.

Makes sense, will be fixed in next version.

> So it's a new API, but where's the user?
> I guess just include this patch with where-ever it's actually used.

The user of virtio_pci_add_shm_cap is in the previous commits.



My original patch [0] was actually a squash of the current two commits, 
but Dr. David Alan Gilbert explicitly asked me [1] to split them in 
order to preserve his original virtio-pci patch [2]. I could squash 
these two commits together again, but we will be back to square one [1].



[0] https://www.mail-archive.com/qemu-devel@nongnu.org/msg826814.html

[1] https://www.mail-archive.com/qemu-devel@nongnu.org/msg826890.html

[2] 
https://gitlab.freedesktop.org/virgl/qemu/-/commit/7fa847fde7143ca2ef5b0a2a13c5f669d3beb195