Doom.
Le dim. 28 févr. 2021 à 03:42, Andy Bennett
<andyjpb@ashurst.eu.org> a écrit :
Hi,
I've been trying to get my JTAGICE3 to work in UPDI mode with an ATTiny1604
(One of the 0-series devices) and avrdude under Linux.
I think it's getting stuck whilst trying to read the signature bytes from
the target.
I'm using the SVN version of avrdude from a few weeks ago.
Initially I had firmware 1.25 on the JTAGICE3 so I had to upgrade it to
3.37 using Atmel Studio under Windows. Otherwise I'm entirely using Linux;
I had to borrow a machine to get the firmware updated.
Whilst I was there I tested that Atmel Studio could use the JTACICE3 with
the new firmware to read out the signature bytes of an ATMega1284p via JTAG
and an ATTiny1604 via UPDI. Both were successful.
The JTAGICE3 with the new firmware can also read the ATMega1284p with
avrdude in JTAG mode under linux.
However, it hangs with the ATTiny1604:
-----
$ ~/project/.../avrdudue/installation/bin/avrdude -p t1604 -c jtag3updi -P
usb -t
avrdude: AVR device initialized and ready to accept instructions
Reading | | 0% 0.00s
-----
It just sits here forever, never making any progress.
If I increase the debugging level and attach GDB I see that it gets stuck
in avr.c:avr_read()'s call to jtag.c:jtag3_paged_load().
jtag3_paged_load() gets stuck in an infinite loop.
It gets passed the correct parameters:
-----
(gdb) bt
#0 0x00007fbd14811ff7 in __GI___select (nfds=4, readfds=0x0,
writefds=0x7ffff003dca0, exceptfds=0x0, timeout=0x7ffff003dc30)
at ../sysdeps/unix/sysv/linux/select.c:41
#1 0x00007fbd14aa57ad in ?? () from /lib/x86_64-linux-gnu/libusb-0.1.so.4
#2 0x0000556ff4bf676d in usbdev_recv_frame (fd=0x556ff5ff6610,
buf=0x556ff62c5880 "\020m,\366oU", nbytes=512)
at usb_libusb.c:506
#3 0x0000556ff4c1560f in jtag3_edbg_recv_frame (msg=0x7ffff003def8,
pgm=0x556ff5ff50d0) at jtag3.c:709
#4 jtag3_recv_frame (msg=0x7ffff003def8, pgm=0x556ff5ff50d0) at
jtag3.c:647
#5 jtag3_recv (pgm=pgm@entry=0x556ff5ff50d0, msg=msg@entry=0x7ffff003def8)
at jtag3.c:780
#6 0x0000556ff4c15bf1 in jtag3_command (pgm=pgm@entry=0x556ff5ff50d0,
cmd=cmd@entry=0x7ffff003df04 "\022!",
cmdlen=cmdlen@entry=12, resp=resp@entry=0x7ffff003def8,
descr=descr@entry=0x556ff4c332bc "read memory") at jtag3.c:827
#7 0x0000556ff4c1677c in jtag3_paged_load (pgm=0x556ff5ff50d0,
p=0x556ff62754e0, m=0x556ff62766f0, page_size=0, addr=0,
n_bytes=1) at jtag3.c:1811
#8 0x0000556ff4be3f07 in avr_read (pgm=pgm@entry=0x556ff5ff50d0,
p=p@entry=0x556ff62754e0,
memtype=memtype@entry=0x556ff4c1c165 "signature", v=v@entry=0x0) at
avr.c:406
#9 0x0000556ff4be4733 in avr_signature (pgm=0x556ff5ff50d0,
p=0x556ff62754e0) at avr.c:1055
#10 0x0000556ff4bdfc1a in main (argc=<optimized out>, argv=<optimized out>)
at main.c:1046
-----
In frame 7, m is the "signature" region with page_size = 1 and readsize =
0.
-----
(gdb) frame 7
#7 0x0000556ff4c1677c in jtag3_paged_load (pgm=0x556ff5ff50d0,
p=0x556ff62754e0, m=0x556ff62766f0, page_size=0, addr=0,
n_bytes=1) at jtag3.c:1811
1811 if ((status = jtag3_command(pgm, cmd, 12, &resp, "read
memory")) < 0)
(gdb) print m
$1 = (AVRMEM *) 0x556ff62766f0
(gdb) print m->readsize
$2 = 0
(gdb) print m->page_size
$3 = 1
(gdb) print page_size
$4 = 0
-----
Then, on line 1769 of jtag3.c , jtag3_paged_load() does
-----
page_size = m->readsize;
-----
...which causes the for loop on lin 1796:
-----
for (; addr < maxaddr; addr += page_size) {
-----
... to go into an infinite loop because m->readsize == 0.
I've tried commenting out the assignment to page_size and I can get avrdude
to go into the interactive terminal:
-----
$ ~/project/.../avrdudue/installation/bin/avrdude -p t1604 -c jtag3updi -P
usb -t
avrdude: AVR device initialized and ready to accept instructions
Reading | | 0%
0.00savrdude: bad response to read memory command: 0xa0
Reading | ################################################## | 100% 0.01s
avrdude: Device signature = 0x1e9425 (probably t1604)
avrdude> ^C
-----
However, my cursory inspection of both the SVN history and the atmelice
driver suggest that this code has been untouched for a very long time and
is therefore most likely not at fault.
I've also scoured avrdude.conf to see if there's anything unusual about how
the ATTiny1604 is set up. I didn't notice anything and no parts seem to
have readsize assigned for the "signature" memory.
I've not looked in detail at how the JTAGICE3 is talking to the ATMega1284p
in JTAG mode but the jtag3.c driver doesn't seem to define a
`read_sig_bytes` function anywhere so it's not falling into that special
case in avr_read (avr.c, line 430). Perhaps it's using the read_byte()
calls on line 440?
So my question is, does anyone know why the JTAGICE3 is hanging like this?
Atmel Studio can successfully connect to the same board using the same
JTAGICE3 and the verbose logging output of avrdude tells me that not only
is it succesfully connecting to the JTAGICE3 and the target, it is also
able to send the AVR sign-on command and read the SIB, so I'm pretty sure
the hardware is all OK. I can also interact with the device using pyudpi on
Linux.
-----
avrdude: Sending AVR sign-on command:
avrdude: jtag3_edbg_send(): sending 4 bytes
avrdude: jtag3_edbg_recv():
avrdude: jtag3_recv(): Got message seqno 14 (command_sequence == 14)
Raw message:
12 84 00 74 69 6e 79 00
[AVR] Data returned:
0x00 0x74 0x69 0x6e 0x79 0x00
avrdude: Partial Family_ID returned: "tiny"
avrdude: AVR device initialized and ready to accept instructions
Reading | avrdude: jtag3_paged_load(.., signature, 1, 0x0, 1)
mapped to address: 0x1100
avrdude: Sending enter progmode command:
avrdude: jtag3_edbg_send(): sending 3 bytes
avrdude: jtag3_edbg_recv():
avrdude: jtag3_recv(): Got message seqno 15 (command_sequence == 15)
-----
I've attached a (compressed) log that shows the output of
-----
$ ~/project/.../avrdudue/installation/bin/avrdude -p t1604 -c jtag3updi -P
usb -t -vvvv > avrdude-jtagice3updi.log 2>&1
-----
Thanks for any pointers you can offer!
Best wishes,
@ndy
avrdude.conf