Porting openwrt(kernel 5.4) for bpi-r2 status report

frank-w · April 13, 2020, 12:51pm

i guess path is wrong…as driver was designed for android the firmware has to be in /system/etc/firmware also WMT_SOC.cfg is needed

see this documentation:

cioby23 · April 13, 2020, 12:53pm

I just checked now and WMT_SOC.cfg is there

BPI-R2> ls mmc 1:2 /system/etc/firmware
<DIR>       4096 .
<DIR>       4096 ..
              95 WMT_SOC.cfg

Do you suggest to move the firmware files also under system folder ? Int he README file it says

copy the following file to BPI-ROOT-partition of SD-Card:

config\WMT_SOC.cfg to system/etc/firmware/
content of folder firmware to etc/firmware/
src\stp_uart_launcher, src\wmt_loader and src\wmt_loopback to usr/bin/

frank-w · April 13, 2020, 12:55pm

no, should be right…my mistake…firmware should be in /etc/firmware

maybe @smaller09 or @abbradar has an idea…i think he used my version, and 5.4 works with debian…but maybe openwrt make some changes that break…

it’s bad, that call stack is not really explaining where the crash happens

TheHippo1555 · April 29, 2020, 7:32pm

hi. i tryied to compile and i always end with error.

/Include/toplevel.mk:255: recipe for target ‘world’ failed.

make: *** [world] Error 1

any ideas what is wrong ?

cioby23 · April 29, 2020, 8:54pm

This is a generic error. Try to compile again with verbose mode enabled.

make -j1 V=s

This will output more information and you’ll be able to see where the compilation is failing.

spikerguy · May 9, 2020, 11:14am

Hello, Any update on the wifi fix?

frank-w · May 9, 2020, 7:22pm

As far as i understand it, crash seems to happen near here:

github.com

frank-w/BPI-R2-4.14/blob/1beb6809a1184a9074f5c4544f093b374fcac74d/drivers/misc/mediatek/btif/common/mtk_btif.c#L1882


		BTIF_WARN_FUNC("Null dev pointer!!!!\n");


	p_vfifo = p_dma->p_dma_info->p_vfifo;
	if (p_vfifo->p_vir_addr != NULL) {
		BTIF_ERR_FUNC
		    ("BTIF vFIFO memory already allocated, do nothing\n");
		return E_BTIF_BAD_POINTER;
	}


/*vFIFO memory allocation*/
	p_vfifo->p_vir_addr = dma_alloc_coherent(dev,
						  p_vfifo->vfifo_size,
						  &p_vfifo->phy_addr, GFP_DMA | GFP_DMA32);
	if (p_vfifo->p_vir_addr == NULL) {
		BTIF_ERR_FUNC("alloc vFIFO memory for BTIF failed\n");
		return E_BTIF_FAIL;
	}


	if (sizeof(dma_addr_t) == sizeof(unsigned long long))
		BTIF_INFO_FUNC("alloc vFIFO for BTIF succeed in arch64,vir addr:0x%p,",
		"phy addr:0x%llx\n", p_vfifo->p_vir_addr, p_vfifo->phy_addr);

Have not seen it in my kernel…maybe because ENABLE_BTIF_TX_DMA is set in yours and not mine…

You can try adding printk-lines to trace exact position of crash with dumping values of pointers before accessing them…i guess p_vfifo->phy_addr is NULL