Re: [PATCH] driver: gadget: fastboot: Link endpoint and descriptors
在 2021/11/4 0:53, Sean Anderson 写道:
On 9/16/21 3:02 AM, qianfanguijin@163.com wrote:
From: qianfan Zhao qianfanguijin@163.com
If the downloading file size is equal to the partition size, "fastboot flash" can't work, at least in sunxi platform, because used an uninitalized point: ep->desc.
Hm, I think that usb_ep_ops->enable needs to set ep->desc = desc on success.
Of the existing drivers, only musb-new and mtu3 skip this.
I checked the udc driver and found that not all the udc driver set "ep->desc = desc", such as atmel_usba_udc, dwc2_udc_otg, musb-new and mtu3. Those drivers save "desc" to bsp driver's private data only, such as "msub_ep->desc = desc", but the common usb_ep's desc is an invalid point.
And I'm not find any documents declare the behaves how usb_ep_ops->enable did, save desc or not. So I'd prefect save desc in the fastboot level, and the udc driver can also save desc again if they want.
--Sean
On 11/5/21 2:28 AM, qianfan wrote:
在 2021/11/4 0:53, Sean Anderson 写道:
On 9/16/21 3:02 AM, qianfanguijin@163.com wrote:
From: qianfan Zhao qianfanguijin@163.com
If the downloading file size is equal to the partition size, "fastboot flash" can't work, at least in sunxi platform, because used an uninitalized point: ep->desc.
Hm, I think that usb_ep_ops->enable needs to set ep->desc = desc on success.
Of the existing drivers, only musb-new and mtu3 skip this.
I checked the udc driver and found that not all the udc driver set "ep->desc = desc", such as atmel_usba_udc, dwc2_udc_otg, musb-new and mtu3. Those drivers save "desc" to bsp driver's private data only, such as "msub_ep->desc = desc", but the common usb_ep's desc is an invalid point.
And I'm not find any documents declare the behaves how usb_ep_ops->enable did, save desc or not. So I'd prefect save desc in the fastboot level, and the udc driver can also save desc again if they want.
This idiom is also present in f_thor.c, f_rockusb.c, and f_mass_storage.c
Given that a majority of otg implementations set this, I think the other should be changed to match.
--Sean
Hi:
I had pushed a new version and you can read it from http://patchwork.ozlabs.org/project/uboot/patch/tencent_AECDF2ED56A768BA5AFB...
This also fixed fastboot data abort bug on am335x platform such as:
=> fastboot 0 musb-hdrc: peripheral reset irq lost! ** Bad device specification mmc bootloader_a ** Couldn't find partition mmc bootloader_a Starting download of 108152 bytes data abort pc : [<9ff9b198>] lr : [<9ff9b195>] reloc pc : [<80830198>] lr : [<80830195>] sp : 9df35f30 ip : 00000000 fp : 00000002 r10: 00000000 r9 : 9df4aeb0 r8 : 00000001 r7 : 9df574c0 r6 : 9df4fea4 r5 : 00000001 r4 : 0001a678 r3 : 00000000 r2 : 9df35f48 r1 : 9df35f48 r0 : 0001a678 Flags: nzCv IRQs off FIQs on Mode SVC_32 (T) Code: f7fd fdaa 6823 1e04 (791d) 795b Resetting CPU ...
resetting ... CCCCCCCC U-Boot SPL 2022.01-rc1-01417-gcf8017ac80 (Nov 15 2021 - 13:35:51 +0800) Trying to boot from USB eth eth0: eth_cpsw, eth1: usb_ether using musb-hdrc, OUT ep1out IN ep1in STATUS ep2in MAC de:ad:be:ef:00:01 HOST MAC de:ad:be:ef:00:00
在 2021/11/5 21:29, Sean Anderson 写道:
On 11/5/21 2:28 AM, qianfan wrote:
在 2021/11/4 0:53, Sean Anderson 写道:
On 9/16/21 3:02 AM, qianfanguijin@163.com wrote:
From: qianfan Zhao qianfanguijin@163.com
If the downloading file size is equal to the partition size, "fastboot flash" can't work, at least in sunxi platform, because used an uninitalized point: ep->desc.
Hm, I think that usb_ep_ops->enable needs to set ep->desc = desc on success.
Of the existing drivers, only musb-new and mtu3 skip this.
I checked the udc driver and found that not all the udc driver set "ep->desc = desc", such as atmel_usba_udc, dwc2_udc_otg, musb-new and mtu3. Those drivers save "desc" to bsp driver's private data only, such as "msub_ep->desc = desc", but the common usb_ep's desc is an invalid point.
And I'm not find any documents declare the behaves how usb_ep_ops->enable did, save desc or not. So I'd prefect save desc in the fastboot level, and the udc driver can also save desc again if they want.
This idiom is also present in f_thor.c, f_rockusb.c, and f_mass_storage.c
Given that a majority of otg implementations set this, I think the other should be changed to match.
--Sean
participants (2)
-
qianfan -
Sean Anderson