On 2018-12-18 18:50, Markus Armbruster wrote:
Thomas Huth <thuth(a)redhat.com> writes:
> It's been marked as deprecated in QEMU v2.6.0 already, so really nobody
> should use the legacy "ivshmem" device anymore (but use ivshmem-plain or
> ivshmem-doorbell instead). Time to remove the deprecated device now.
>
> Signed-off-by: Thomas Huth <thuth(a)redhat.com>
> ---
> docs/specs/ivshmem-spec.txt | 8 +-
> hw/i386/pc_piix.c | 4 -
> hw/misc/ivshmem.c | 206 +-------------------------------------------
> qemu-deprecated.texi | 5 --
> scripts/device-crash-test | 1 -
> tests/ivshmem-test.c | 65 +++++---------
> 6 files changed, 29 insertions(+), 260 deletions(-)
>
> diff --git a/docs/specs/ivshmem-spec.txt b/docs/specs/ivshmem-spec.txt
> index a1f5499..042f7ea 100644
> --- a/docs/specs/ivshmem-spec.txt
> +++ b/docs/specs/ivshmem-spec.txt
> @@ -17,12 +17,16 @@ get interrupted by its peers.
>
> There are two basic configurations:
>
> -- Just shared memory: -device ivshmem-plain,memdev=HMB,...
> +- Just shared memory:
> +
> + -device ivshmem-plain,memdev=HMB,...
>
> This uses host memory backend HMB. It should have option "share"
> set.
>
> -- Shared memory plus interrupts: -device ivshmem,chardev=CHR,vectors=N,...
> +- Shared memory plus interrupts:
> +
> + -device ivshmem-doorbell,chardev=CHR,vectors=N,...
>
> An ivshmem server must already be running on the host. The device
> connects to the server's UNIX domain socket via character device
Just whitespace. Intentional?
It's not just whitespace, I had to change "-device ivshmem" into
"-device ivshmem-doorbell" here, and that did not fit into one line anymore.
> @@ -882,8 +864,8 @@ static void ivshmem_common_realize(PCIDevice
*dev, Error **errp)
> IVShmemState *s = IVSHMEM_COMMON(dev);
> Error *err = NULL;
> uint8_t *pci_conf;
> - uint8_t attr = PCI_BASE_ADDRESS_SPACE_MEMORY |
> - PCI_BASE_ADDRESS_MEM_PREFETCH;
> + const uint8_t attr = PCI_BASE_ADDRESS_SPACE_MEMORY |
> + PCI_BASE_ADDRESS_MEM_PREFETCH | PCI_BASE_ADDRESS_MEM_TYPE_64;
> Error *local_err = NULL;
>
> /* IRQFD requires MSI */
Sure this belongs to this patch?
Yes. I had to remove the "if (s->not_legacy_32bit)" below, so I moved
the PCI_BASE_ADDRESS_MEM_TYPE_64 to the other bits above. I could also
leave it below, but I think that would look a little bit lonely there
without the if-statement.
> @@ -903,10 +885,6 @@ static void ivshmem_common_realize(PCIDevice
*dev, Error **errp)
> pci_register_bar(dev, 0, PCI_BASE_ADDRESS_SPACE_MEMORY,
> &s->ivshmem_mmio);
>
> - if (s->not_legacy_32bit) {
> - attr |= PCI_BASE_ADDRESS_MEM_TYPE_64;
> - }
> -
> if (s->hostmem != NULL) {
> IVSHMEM_DPRINTF("using hostmem\n");
>
[...]
> diff --git a/tests/ivshmem-test.c b/tests/ivshmem-test.c
> index c37b196..9811d66 100644
> --- a/tests/ivshmem-test.c
> +++ b/tests/ivshmem-test.c
> @@ -305,20 +305,18 @@ static void *server_thread(void *data)
> return NULL;
> }
>
> -static void setup_vm_with_server(IVState *s, int nvectors, bool msi)
> +static void setup_vm_with_server(IVState *s, int nvectors)
> {
> - char *cmd = g_strdup_printf("-chardev socket,id=chr0,path=%s,nowait "
> - "-device
ivshmem%s,chardev=chr0,vectors=%d",
> - tmpserver,
> - msi ? "-doorbell" :
",size=1M,msi=off",
> - nvectors);
> + char *cmd = g_strdup_printf("-chardev socket,id=chr0,path=%s,nowait
-device"
Awkward line break.
> + "
ivshmem-doorbell,chardev=chr0,vectors=%d",
> + tmpserver, nvectors);
Suggest
char *cmd = g_strdup_printf("-chardev socket,id=chr0,path=%s,nowait "
"-device ivshmem-doorbell,chardev=chr0,vectors=%d",
tmpserver, nvectors);
Ok, I can do that in v2.
Thomas