On 2/24/21 12:40 PM, Stefan Hajnoczi wrote:
> On Tue, Feb 23, 2021 at 05:35:20PM +0000, Sai Pavan Boddu wrote:
>> Hi Philippe,
>>
>>> -----Original Message-----
>>> From: Philippe Mathieu-Daudé <phi...@redhat.com>
>>> Sent: Monday, February 22, 2021 5:34 PM
>>> To: Sai Pavan Boddu <saip...@xilinx.com>; Markus Armbruster
>>> <arm...@redhat.com>; Kevin Wolf <kw...@redhat.com>; Max Reitz
>>> <mre...@redhat.com>; Vladimir Sementsov-Ogievskiy
>>> <vsement...@virtuozzo.com>; Eric Blake <ebl...@redhat.com>; Joel Stanley
>>> <j...@jms.id.au>; Cédric Le Goater <c...@kaod.org>; Vincent Palatin
>>> <vpala...@chromium.org>; Dr. David Alan Gilbert <dgilb...@redhat.com>;
>>> Thomas Huth <th...@redhat.com>; Stefan Hajnoczi <stefa...@redhat.com>;
>>> Peter Maydell <peter.mayd...@linaro.org>; Alistair Francis
>>> <alistair.fran...@wdc.com>; Edgar Iglesias <edg...@xilinx.com>; Luc Michel
>>> <luc.mic...@greensocs.com>; Paolo Bonzini <pbonz...@redhat.com>
>>> Cc: Sai Pavan Boddu <saip...@xilinx.com>; qemu-devel@nongnu.org; qemu-
>>> bl...@nongnu.org
>>> Subject: Re: [PATCH v2 01/22] block: add eMMC block device type
>>>
>>> On 2/22/21 9:20 AM, Sai Pavan Boddu wrote:
>>>> From: Vincent Palatin <vpala...@chromium.org>
>>>>
>>>> Add new block device type.
>>>>
>>>> Signed-off-by: Vincent Palatin <vpala...@chromium.org>
>>>> [SPB: Rebased over 5.1 version]
>>>> Signed-off-by: Sai Pavan Boddu <sai.pavan.bo...@xilinx.com>
>>>> Signed-off-by: Joel Stanley <j...@jms.id.au>
>>>> Signed-off-by: Cédric Le Goater <c...@kaod.org>
>>>> Reviewed-by: Alistair Francis <alistair.fran...@wdc.com>
>>>> ---
>>>> include/sysemu/blockdev.h | 1 +
>>>> blockdev.c | 1 +
>>>> 2 files changed, 2 insertions(+)
>>>>
>>>> diff --git a/include/sysemu/blockdev.h b/include/sysemu/blockdev.h
>>>> index 3b5fcda..eefae9f 100644
>>>> --- a/include/sysemu/blockdev.h
>>>> +++ b/include/sysemu/blockdev.h
>>>> @@ -24,6 +24,7 @@ typedef enum {
>>>> */
>>>> IF_NONE = 0,
>>>> IF_IDE, IF_SCSI, IF_FLOPPY, IF_PFLASH, IF_MTD, IF_SD, IF_VIRTIO,
>>>> IF_XEN,
>>>> + IF_EMMC,
>>>> IF_COUNT
>>>> } BlockInterfaceType;
>>>>
>>>> diff --git a/blockdev.c b/blockdev.c
>>>> index cd438e6..390d43c 100644
>>>> --- a/blockdev.c
>>>> +++ b/blockdev.c
>>>> @@ -83,6 +83,7 @@ static const char *const if_name[IF_COUNT] = {
>>>> [IF_SD] = "sd",
>>>> [IF_VIRTIO] = "virtio",
>>>> [IF_XEN] = "xen",
>>>> + [IF_EMMC] = "emmc",
>>>> };
>>>
>>> We don't need to introduce support for the legacy -drive magic.
>>>
>>> -device should be enough for this device, right?
>> [Sai Pavan Boddu] I was seeing to use -device for emmc. But I see we anyway
>> need blockdev support for this, which would require us the use -drive.
>>
>> Can you give some pointers, how to approach this ?
>
> It is probably not necessary to add a new IF_ constant. Would this work:
>
> -drive if=none,id=emmc0,file=test.img,format=raw
> -device emmc,...,drive=emmc0
>
> Or the more modern:
>
> -blockdev node-name=emmc0,driver=file,filename=test.img
> -device emmc,...,drive=emmc0
>
> ?
>
> (The syntax might need small tweaks but is shows the general idea.)
Yes. This is better.
We could have an "emmc" device inheriting from "sd-card". The "emmc"
property would not be necessary anymore and may be, we could cleanup
up some parts doing :
if (sd->emmc) { /* eMMC */
...
} else {
}
with SDCardClass handlers. the SWITCH_FUNCTION command is a good
candidate, CMD8 also.
C.
