Re: [PATCH v7 3/5] usb: host: ohci-sm501: init genalloc for local memory

[Guenter Roeck]

On Tue, Jun 11, 2019 at 07:26:54PM +0200, Fredrik Noring wrote:
> Hi Guenter,
> 
> > > This patch results in usb access failures when trying to boot from the
> > > sm501-usb controller on sh4 with qemu.
> > > 
> > > usb 1-2.1: reset full-speed USB device number 4 using sm501-usb
> > > sd 1:0:0:0: [sda] tag#0 UNKNOWN(0x2003) Result: hostbyte=0x03 driverbyte=0x00
> > > sd 1:0:0:0: [sda] tag#0 CDB: opcode=0x28 28 00 00 00 08 7c 00 00 f0 00
> > > print_req_error: I/O error, dev sda, sector 2172 flags 80700
> > > usb 1-2.1: reset full-speed USB device number 4 using sm501-usb
> > > sd 1:0:0:0: [sda] tag#0 UNKNOWN(0x2003) Result: hostbyte=0x03 driverbyte=0x00
> > > sd 1:0:0:0: [sda] tag#0 CDB: opcode=0x28 28 00 00 00 01 da 00 00 f0 00
> > > print_req_error: I/O error, dev sda, sector 474 flags 84700
> > > usb 1-2.1: reset full-speed USB device number 4 using sm501-usb
> > > sd 1:0:0:0: [sda] tag#0 UNKNOWN(0x2003) Result: hostbyte=0x03 driverbyte=0x00
> > > sd 1:0:0:0: [sda] tag#0 CDB: opcode=0x28 28 00 00 00 02 da 00 00 f0 00
> > > print_req_error: I/O error, dev sda, sector 730 flags 84700
> > > usb 1-2.1: reset full-speed USB device number 4 using sm501-usb
> > > sd 1:0:0:0: [sda] tag#0 UNKNOWN(0x2003) Result: hostbyte=0x03 driverbyte=0x00
> > > sd 1:0:0:0: [sda] tag#0 CDB: opcode=0x28 28 00 00 00 0b 50 00 00 f0 00
> > > print_req_error: I/O error, dev sda, sector 2896 flags 84700
> > > 
> > > Qemu command line is:
> > > 
> > > The qemu command line is:
> > > 
> > > qemu-system-sh4 -M r2d \
> > >         -kernel ./arch/sh/boot/zImage \
> > > 	-snapshot \
> > > 	-usb -device usb-storage,drive=d0 \
> > > 	-drive file=rootfs.ext2,if=none,id=d0,format=raw \
> > > 	-append 'panic=-1 slub_debug=FZPUA root=/dev/sda rootwait console=ttySC1,115200 earlycon=scif,mmio16,0xffe80000 noiotrap' \
> > > 	-serial null -serial stdio \
> > > 	-net nic,model=rtl8139 -net user -nographic -monitor null
> > > 
> > > Reverting this patch as well as "USB: drop HCD_LOCAL_MEM flag" fixes the
> > > problem. Reverting "USB: drop HCD_LOCAL_MEM flag" alone does not help.
> > > 
> > 
> > This problem is still seen in next-20190611.
> > Has anyone actually tested this code ?
> 
> I tested patches 1, 2 and 5 with v5.0.19. Perhaps yet another part of the
> OHCI subsystem allocates memory from the wrong pool? With some luck it is
> relatively easy to trace backwards from the error messages to the point
> where the memory is being allocated. One way to establish this is to
> sprinkle printk around if-statements. There may be 10-20 levels of calls
> including one or two indirect calls via pointers. Would you be able to do
> that?
> 

I don't think I'll have time to do that anytime soon. Not that I know what
exactly to look for in the first place.

Can you do that debugging yourself ? All you would need is a cross-compiler
(eg from kernel.org), qemu, and a working configuration (the root file
system doesn't really matter since the code doesn't get to the point of
loading it, but you can use [1]). For the configuration file, you can use
rts7751r2dplus_defconfig with CONFIG_CMDLINE and CONFIG_CMDLINE_OVERWRITE
removed.

Thanks,
Guenter

---
[1] https://github.com/groeck/linux-build-test/blob/master/rootfs/sh/rootfs.ext2.gz