Re: [PATCH] hw/ssi: imx_spi: Improve chip select handling

[Guenter Roeck]

On 9/8/21 2:05 AM, Cheng, Xuzhou wrote:
> Thanks Bin added me into this loop.
>
> Hi, Guenter
>
> I am interested in your patch and the issue what you found. I want to reproduce 
> your issue on Linux, but I failed, the spi-nor of sabrelite on Linux does not 
> work.
>
> Could you share your Linux kernel version? It would be great if you can share 
> the detailed steps to reproduce.
>
> My test record:
> Linux version: https://github.com/torvalds/linux, the last commit is 
> ac08b1c68d1b1ed3cebb218fc3ea2c07484eb07d.
> Linux configuration: imx_v6_v7_defconfig.
>
> QEMU command:
> qemu-system-arm -nographic -M sabrelite -smp 4 -m 1G -serial null -serial /dev/pts/50 
> -kernel zImage -initrd rootfs.ext4 -append "root=/dev/ram rdinit=sbin/init" 
> -dtb imx6q-sabrelite.dtb -net nic -net tap,ifname=tap_xcheng,script=no -drive 
> file=flash.sabre,format=raw,if=mtd
>
> Logs: there are same logs when I apply your patch or not apply. So I don't 
> understand what this patch fixes, maybe I missed something? :(...
>
> # cat /proc/mtd
> dev:    size   erasesize  name
> mtd0: 00200000 00001000 "spi0.0"
> # ls /dev/mtd*
> /dev/mtd0       /dev/mtd0ro     /dev/mtdblock0
> # echo "01234567899876543210" > test
> # dd if=test of=/dev/mtd0  /* flash.sabre is empty */
> 0+1 records in
> 0+1 records out
> # dd if=/dev/mtd0 of=test_out
> 4096+0 records in
> 4096+0 records out
> # cat test_out             /* test_out is empty */

It is a flash. I don't think dd erases the flash, so unless your flash.sabre
is all-ones you probably won't see the overwritten data.

Guenter