Error when trying to launch gdb server for LEON3FT (GR712RC hard-core CPU)

[robotron]

Dear Patrick,

first of all, thank you for starting your free LEON gdb support
project. Today, I tried to use it together with our fresh GR712RC
board (containing hardware, dual-core LEON3FT CPU). For this CPU,
there is no grmon-eval, so I tried your playtag at the first place.

According to chip's documentation, there seems to be pretty same JTAG-
AHB master bridge (two registers, 33 and 35 bit). I have been
successful while writing to some timer peripheral register using
UrJTAG as follows:

instruction length 6

register DBGCMD 35
register DBGDATA 33

instruction s_cmd 000010 DBGCMD
instruction s_data 000011 DBGDATA

ins s_cmd
shift ir
dr 11010000000000000000000001100000100
shift dr
dr
ins s_data
shift ir
dr 010111010101100011101111011010000
shift dr
dr
#
ins s_cmd
shift ir
dr 01010000000000000000000001100000100
shift dr
dr
ins s_data
shift ir
shift dr
dr
jtag> jtag> 11010000000000000000000001100000100 (0x680000304)
jtag> jtag> 01010000000000000000000001100000100 (0x280000304)
jtag> jtag> jtag> 010111010101100011101111011010000 (0xBAB1DED0)
jtag> jtag> 010111010101100011101111011010000 (0xBAB1DED0)
jtag> jtag> jtag> jtag> 01010000000000000000000001100000100
(0x280000304)
jtag> jtag> 11010000000000000000000001100000100 (0x680000304)
jtag> jtag> jtag> jtag> 000000000000000001101111011010000 (0x0000DED0)
jtag>

...so it seems to me that this lowest-level interface works.

Then, I compared your dsuregs.py with the docs very briefly and it
seems the DSU registers are the same in the LEON3FT variant, too.


However, while trying to run ./leonserver.py, I got following error:

JTAG Chain information

#0 - Gaisler Research AB gr712rc (ir_capture = '001001',
idcode='0000_0000011100010010_01100100100_1')

Error opening cable driver (details below)

(...something...)

Traceback (most recent call last):
File "./leonserver.py", line 33, in <module>
processor = CmdProcessor(LeonMem(driver, config), config)
File "./../../playtag/leon3/gdbproc.py", line 159, in __init__
self.leoncfg = leoncfg = LeonCfg(ahb, userconfig)
File "./../../playtag/leon3/cpustate.py", line 74, in __init__
self.calcram()
File "./../../playtag/leon3/cpustate.py", line 85, in calcram
blocksize = min(x.ram_size for x in cfgs)
ValueError: min() arg is an empty sequence

---
my files:

* leongdb.cfg:

# This is the example configuration file.

# Modify it and save it as leongdb.cfg

JTAGID_FILE = "jtagid.txt" # Please use forward slashes...

DSU_ADDR = 0x90000000 # Location of DSU in AHB space

DSU_ENABLE_ADDR = 0x90000000 # Location of register to write to
# to enable DSU (0 for none)
DSU_ENABLE_DATA = 0x400 # Value to write to DSU_ENABLE_ADDR

AHB_RAM_ADDR = 0x0a0000000 # Location of program RAM in AHB
space
AHB_RAM_SIZE = 0x00030000 # Size of AHB RAM

if 1:
CABLE_DRIVER = "ftdi"
CABLE_NAME = 0
else:
CABLE_DRIVER = "digilent"
CABLE_NAME = "DCabUsb"

* jtagid.txt:

# Device Name Device ID Device ID Mask Instr. Reg.
Length Dbg. Interface Instr. 1 Instr. 2
gr712rc 0x00712649 0x0fffffff 6 1 0x2 0x3


Please, do you have and idea, what am I doing wrong?

Thank you,
Marek

[Patrick Maupin]

Marek:

I'm not sure, but at first glance I wonder if my algorithm is very
robust in the face of non power-of-two RAM sizes. Try reducing your
RAM size to 0x20000. Meanwhile I'll try to look at the code to see
what else could cause this.

Best regards,
Pat

[Patrick Maupin]

Oh, never mind. It's a silly bug where it expects to find some TCMs
-- nothing to do with your other RAM.

It uses the TCM info because TCM (tightly coupled memories) aren't at
that AMBA address and it has to do a translation. I'll study it a bit
and try to come up with a fix.

Pat

[robotron]

Dear Patrick,

> Oh, never mind. It's a silly bug where it expects to find some TCMs
> -- nothing to do with your other RAM.
>
> It uses the TCM info because TCM (tightly coupled memories) aren't at
> that AMBA address and it has to do a translation. I'll study it a bit
> and try to come up with a fix.

thank you for your very quick reply. I am looking forward to test it after
your fix. To save your time, if you can tell me where and what to hotfix,
I could try...

Greetings,
Marek

[Patrick Maupin]

Looks like the fix for when you don't have tightly coupled memories
might be fairly simple. See if replacing this line works:

- blocksize = min(x.ram_size for x in cfgs)
+ blocksize = cfgs and min(x.ram_size for x in cfgs) or 2

On Feb 14, 1:35 pm, robotron <hefais...@gmail.com> wrote:

[hefa...@gmail.com]

> - blocksize = min(x.ram_size for x in cfgs)
> + blocksize = cfgs and min(x.ram_size for x in cfgs) or 2

Great, now it is:
Waiting for connection on 0.0.0.0:2222 (Ctrl-C to exit)

So I will setup the toolchain and I hope I will be able to confirm you
functionality, soon.

Thank you for your good work and support!
Marek

[hefa...@gmail.com]

Dear Patrick,

now trying with bare gdb using your ./leonserver.py
...and it seems it really does something! Excellent...

(gdb) set *0x80000304 = 0xbab1ded0
Current language: auto; currently asm
(gdb) x 0x80000304
0x80000304: 0x0000ded0

Many thanks and greetings,
Marek

[Patrick Maupin]

You're welcome!

Glad you could get it to work.

Regards,
Pat