On Tue, Dec 31, 2019 at 11:18 PM Heinrich Schuchardt xypron.glpk@gmx.de wrote:
CC Bin Meng
On 12/31/19 4:08 PM, Heinrich Schuchardt wrote:
On 12/31/19 2:31 PM, Sughosh Ganu wrote:
Scan the pci bus in board_init routine before scanning the virtio devices. This enumerates all the virtio devices, including devices found on the pci bus.
Signed-off-by: Sughosh Ganu sughosh.ganu@linaro.org
board/emulation/qemu-arm/qemu-arm.c | 7 +++++++ 1 file changed, 7 insertions(+)
diff --git a/board/emulation/qemu-arm/qemu-arm.c b/board/emulation/qemu-arm/qemu-arm.c index 4e18733..6c5335c 100644 --- a/board/emulation/qemu-arm/qemu-arm.c +++ b/board/emulation/qemu-arm/qemu-arm.c @@ -63,6 +63,13 @@ struct mm_region *mem_map = qemu_arm64_mem_map; int board_init(void) { /*
* Scan the pci bus before calling virtio_init. This* enumerates all virtio devices, including devices* on the pci bus.The last sentence is easy to misunderstand. You surely do not mean that pci_init() would enumerate all virtio devices including mmio virtio devices.
I would suggest to rephrase it to "This enumerates all pci devices including virtio devices on the pci bus."
Sounds good.
With the patch I no longer need to run `virtio scan` on qemu_arm64_defonfig before using the rng command supplied in patch
[PATCH 1/1] cmd: add rng command https://lists.denx.de/pipermail/u-boot/2019-December/394539.html
Unfortunately you only considered the ARM architecture. We need a solution that works on all QEMU architectures (MIPS, RISC-V, X86, ARM).
Agreed, sorry I missed that. Actually all the board_init() look similar so we should do the same.
How about eliminating board_init() for ARM and RISC-V and moving virtio_init() to common/board_r.c after initr_pci?
Or maybe we introduce a new driver model flag so that uclass drivers are automatically probed?
Regards, Bin