search

litex-hub / linux-on-litex-vexriscv

Linux on LiteX-VexRiscv
BSD 2-Clause "Simplified" License
574 stars 174 forks source link

OpenSBI hangs during serial boot on Arty A7 #206

Closed hasheddan closed 3 years ago

hasheddan commented 3 years ago

After loading bitstream (/make.py --board=arty --load), when serial booting (sudo lxterm --images=images/boot.json /dev/ttyUSB1 --serial-boot --speed=1e6) on Digilent Arty A7 OpenSBI appears to hang forever on sit: IPv6, IPv4 and MPLS over IPv4 tunneling driver. Tried it out with both kernel versions from #164. Let me know if there is any additional information I can provide. Thanks for y'all's awesome work on this!

--============= Liftoff! ===============--

OpenSBI v0.8-1-gecf7701
   ____                    _____ ____ _____
  / __ \                  / ____|  _ \_   _|
 | |  | |_ __   ___ _ __ | (___ | |_) || |
 | |  | | '_ \ / _ \ '_ \ \___ \|  _ < | |
 | |__| | |_) |  __/ | | |____) | |_) || |_
  \____/| .__/ \___|_| |_|_____/|____/_____|
        | |
        |_|

Platform Name       : LiteX / VexRiscv-SMP
Platform Features   : timer,mfdeleg
Platform HART Count : 8
Boot HART ID        : 0
Boot HART ISA       : rv32imas
BOOT HART Features  : time
BOOT HART PMP Count : 0
Firmware Base       : 0x40f00000
Firmware Size       : 124 KB
Runtime SBI Version : 0.2

MIDELEG : 0x00000222
MEDELEG : 0x0000b101
[    0.000000] Linux version 5.11.0-rc2 (florent@panda) (riscv32-buildroot-linux-gnu-gcc.br_real (Buildroot 2020.11-281-g69e5046e7b) 10.2.0, GNU ld (GNU Binutils) 2.33.1) #2 SMP Mon Jan 11 10:05:20 CET 2021

...
[    1.606276] LiteX SoC Controller driver initialized: subreg:4, align:4
[    2.597710] f0001000.serial: ttyLXU0 at MMIO 0x0 (irq = 0, base_baud = 0) is a liteuart
[    2.602984] printk: console [liteuart0] enabled
[    2.602984] printk: console [liteuart0] enabled
[    2.605283] printk: bootconsole [sbi0] disabled
[    2.605283] printk: bootconsole [sbi0] disabled
[    2.623210] litex-spiflash f0006000.spiflash: s25fl128s1 (16384 Kbytes)
[    2.665384] libphy: Fixed MDIO Bus: probed
[    2.678509] liteeth f0004800.mac eth0: irq 2, mapped at a0415800
[    2.683099] i2c /dev entries driver
[    2.689960] i2c i2c-0: Not I2C compliant: can't read SCL
[    2.690861] i2c i2c-0: Bus may be unreliable
[    2.720638] litex-mmc f0006800.mmc: Requested clk_freq=12500000: set to 12500000 via div=8
[    2.749172] fpga_manager fpga0: LiteX ICAPBitstream FPGA Manager registered
[    2.773361] NET: Registered protocol family 10
[    2.788482] Segment Routing with IPv6
[    2.790187] sit: IPv6, IPv4 and MPLS over IPv4 tunneling driver
Dolu1990 commented 3 years ago

Hi,

Does it hang systematicaly at the exact same point ?

About the hang, it seems to me it is in the kernel boot, not opensbi ^^

Maybe it is related to something broken in liteeth ? Can you try a boot without liteeth enabled ?

hasheddan commented 3 years ago

@Dolu1990 thanks for following up! Yes it hangs at the exact same point every time. To boot without liteeth, should I just disable it here by setting to n and rebuild the linux images?

Dolu1990 commented 3 years ago

About litheth, i think there is an option into the make.py command line isn't it ? check make.py --help (sorry i'm in travel, i can't check exactly)

This should be simpler than rebuilding a whole linux image.

hasheddan commented 3 years ago

@Dolu1990 no worries at all, thank you for your help! I don't see a flag, but I'm guessing commenting out the ethernet capability would have the desired effect:

https://github.com/litex-hub/linux-on-litex-vexriscv/blob/7dde41de91aee99a52337b2deebc31c91b7ce862/make.py#L58

hasheddan commented 3 years ago

Hmm that doesn't seem to change anything... Interestingly, if I comment out the ethernet line, then load the bitstream, then serial boot the LEDs stop flashing at the point the boot hangs. This doesn't happen when ethernet line is not commented out :thinking:

geertu commented 3 years ago

Your issue looks very similar to #176. However, the images you used should contain the related fixes.

hasheddan commented 3 years ago

@geertu is it possible the images were not updated after the fix went in? (https://github.com/litex-hub/linux-on-litex-vexriscv/issues/176#issuecomment-757876329)

enjoy-digital commented 3 years ago

@hasheddan: Just to confirm, commenting out ethernetcapability in the target will disable Ethernet (both in the gateware and linux (through the .dts)). I'm going to rebuild and test the Arty target with prebuilt linux image.

enjoy-digital commented 3 years ago

@hasheddan: I don't see the issue on my setup with a fresh bitstream and using prebuilt OpenSBI/Linux images. Can you test with the bitstream/images attached?

I just loaded the bistream and then the images with: litex_term /dev/ttyUSBX --speed=1e6 --images=images/boot.json

arty_2021_03_11.zip images_2021_03_11.zip

hasheddan commented 3 years ago

@enjoy-digital thanks for following up! I'll give those a try. Do they match the ones on #164?

hasheddan commented 3 years ago

@enjoy-digital those artifacts work! I wonder if the ones listed on #164 are stale?

arty-litex

enjoy-digital commented 3 years ago

@hasheddan: Good, to be sure I reused the Linux/OpenSBI images from #164 (but rebuilt the bitstream). So the difference with your previous tests is the bistream and rv32.dtb.

enjoy-digital commented 3 years ago

Not sure what happened exactly, but I think we can close this since has been tested sucesfully with the re-generated bitstream and prebuilt images.