ESP8266 Community Forum

- Sat Jun 06, 2015 3:30 pm #19677 Moin,

some progress:

A session using gdb from repo of @jcmvbkbc plus prologue hack on some instrumented nodemcu:

Booting the firmware:

Code: Select alluser_init done
nodemcu_init

NodeMCU 0.9.5 build 20150606 (dev branch 20150604, pcall protect, ex catch, cal_gdbstub)  powered by Lua 5.1.4
lua: cannot open init.lua
>


Doing some lua and setup a timer that gets shot after some time and calls a BREAK.

Code: Select all> =2+3*4
14
> tmr.alarm(1,10000,0, function() node.breakpoint() end)
> =1+2+3+4+5
15
>
  $T05#b9


OK, leaving terminal.

.gdbinit:

Code: Select allfile app/.output/eagle/debug/image/eagle.app.v6.out
#set debug remote 1
# set a serial line BREAK to remote when typing CTRL-c on gdb prompt.
set remote interrupt-sequence BREAK
#set remote interrupt-on-connect 1
target remote /dev/ttyUSB0
#set remotebaud 115200
set serial baud 115200
add-symbol-file ../esp-elf-rom/esp-elf-rom/bootrom.elf 0x40000000


Code: Select all../../jcmvbkbc/binutils-gdb-xtensa/gdb/gdb


gives

Code: Select allGNU gdb (binutils-gdb-xtensa) 7.9.50.20150529-cvs
Copyright (C) 2015 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
and "show warranty" for details.
This GDB was configured as "--host=x86_64-build_unknown-linux-gnu --target=xtensa-lx106-elf".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>.
Find the GDB manual and other documentation resources online at:
<http://www.gnu.org/software/gdb/documentation/>.
For help, type "help".
Type "apropos word" to search for commands related to "word".
gdb_breakpoint () at arch_esp8266.c:35
35      asm("break 0,0");
add symbol table from file "../esp-elf-rom/esp-elf-rom/bootrom.elf" at
   .text_addr = 0x40000000
warning: section .text not found in .../esp-elf-rom/esp-elf-rom/bootrom.elf
(gdb)


backtrace:

Code: Select all(gdb) bt
#0  gdb_breakpoint () at arch_esp8266.c:35
#1  0x40254b28 in node_breakpoint (L=<optimized out>) at node.c:33
#2  0x40244583 in luaD_precall (L=L@entry=0x3fff7920, func=<optimized out>, func@entry=0x3fff7b78, nresults=nresults@entry=0) at ldo.c:340
#3  0x4024ae86 in luaV_execute (L=L@entry=0x3fff7920, nexeccalls=nexeccalls@entry=1) at lvm.c:646
#4  0x402446c9 in luaD_call (L=L@entry=0x3fff7920, func=0x3fff7b70, nResults=-1) at ldo.c:430
#5  0x40241245 in f_call (L=L@entry=0x3fff7920, ud=ud@entry=0x3fffff00) at lapi.c:855
#6  0x40243db8 in luaD_rawrunprotected (L=L@entry=0x3fff7920, f=f@entry=0x40241234 <f_call>, ud=ud@entry=0x3fffff00) at ldo.c:123
#7  0x402447f8 in luaD_pcall (L=L@entry=0x3fff7920, func=func@entry=0x40241234 <f_call>, u=u@entry=0x3fffff00, old_top=<optimized out>, ef=<optimized out>) at ldo.c:519
#8  0x40241ee1 in lua_pcall (L=L@entry=0x3fff7920, nargs=nargs@entry=0, nresults=-1, errfunc=errfunc@entry=1) at lapi.c:877
#9  0x40240bfe in cal_ido_lua_call (L=L@entry=0x3fff7920, narg=narg@entry=0, clear=clear@entry=0, line=line@entry=0x3ffeb364 "24") at lua.c:160
#10 0x40241028 in cal_do_lua_call (L=L@entry=0x3fff7920, narg=narg@entry=0, clear=clear@entry=0, line=line@entry=0x3ffeb364 "24") at cal_debug.c:58
#11 0x40255b84 in alarm_timer_common (L=0x3fff7920, id=id@entry=1) at tmr.c:24
#12 0x40255bb0 in alarm_timer_cb1 (arg=<optimized out>) at tmr.c:36
#13 0x40240d49 in func2 (data=<optimized out>) at cal_debug.c:74
#14 0x40002e09 in ets_timer_handler_isr ()
#15 0x40002e51 in ets_timer_handler_isr ()
#16 0x40000f49 in ets_post ()


Some navigation:

Code: Select all(gdb) up 2
#2  0x40244583 in luaD_precall (L=L@entry=0x3fff7920, func=<optimized out>, func@entry=0x3fff7b78, nresults=nresults@entry=0) at ldo.c:340
340         n = ((lua_CFunction)fvalue(ci->func))(L);  /* do the actual call */
(gdb) list
335         luaD_callhook(L, LUA_HOOKCALL, -1);
336       lua_unlock(L);
337       if (ttisfunction(ci->func))
338         n = (*curr_func(L)->c.f)(L);  /* do the actual call */
339       else
340         n = ((lua_CFunction)fvalue(ci->func))(L);  /* do the actual call */
341       lua_lock(L);
342       if (n < 0)  /* yielding? */
343         return PCRYIELD;
344       else {
(gdb) print ci
$1 = (CallInfo *) 0x3fff7ac0
(gdb) print *ci
$2 = {base = 0x3fff7b80, func = 0x3fff7b78, top = 0x3fff7c20, savedpc = 0x3fff8aac, nresults = 0, tailcalls = -625713182}
(gdb) print ci->func
$3 = (StkId) 0x3fff7b78
(gdb) print L
$4 = (lua_State *) 0x3fff7920
(gdb) print *L
$5 = {next = 0x0, tt = 10 '\n', marked = 97 'a', status = 0 '\000', top = 0x3fff7b80, base = 0x3fff7b80, l_G = 0x3fff7988, ci = 0x3fff7ac0, savedpc = 0x3fff8e0c, stack_last = 0x3fff7c98, stack = 0x3fff7b60,
  end_ci = 0x3fff7b38, base_ci = 0x3fff7a90, stacksize = 45, size_ci = 8, nCcalls = 1, baseCcalls = 0, hookmask = 0 '\000', allowhook = 1 '\001', basehookcount = 0, hookcount = 0, hook = 0x0, l_gt = {value = {
      gc = 0x3fff7cd8, p = 0x3fff7cd8, n = 1073708248, b = 1073708248}, tt = 7}, env = {value = {gc = 0x3fff7cd8, p = 0x3fff7cd8, n = 1073708248, b = 1073708248}, tt = 7}, openupval = 0x0, gclist = 0x0,
  errorJmp = 0x3ffffe90, errfunc = 8}


Thanks for your help!

Any chance to get a more current gdb into esp8266 esp-open-sdk and/or crosstools ng?

Cal

Get your modules here:

https://www.banggood.com/custlink/KvGGhGF4wG

- Sat Jun 06, 2015 3:51 pm #19681

Any chance to get a more current gdb into esp8266 esp-open-sdk and/or crosstools ng?

I'm currently updating to crosstool-NG 1.21.0 with binutils-2.25, gcc-5.1 and gdb-7.9.1. Expecting to push new branch tomorrow.

- Sat Jun 06, 2015 3:59 pm #19683

Any chance to get a more current gdb into esp8266 esp-open-sdk and/or crosstools ng?

I'm currently updating to crosstool-NG 1.21.0 with binutils-2.25, gcc-5.1 and gdb-7.9.1. Expecting to push new branch tomorrow.

Perfect timing!

Does that contain the a0_base fix?
Could you find some solution to the *end of prologue* detection problem?
Or can we at least add some offset hack for the mean time?

I used

Code: Select allif (prologue_sal.line != 0) /* Found debug info.  */
   body_pc = prologue_sal.end + 40


Cal

Get your modules here:

https://www.banggood.com/custlink/KvGGhGF4wG

- Sat Jun 06, 2015 4:05 pm #19684

Does that contain the a0_base fix?
Could you find some solution to the *end of prologue* detection problem?
Or can we at least add some offset hack for the mean time?

It's only version bump, I haven't done anything for the gdb itself. Can put temporary fixes to local-patches for now.