I have recently started seeing FPC internal error 200310221 when compiling a changed program with lazarus using the trunk fpc version. If I do a build, the problem goes away, only to recur after a random number of changes/recompiles. The program I am working on is quite large, so it is not obvious how to generate a simple test case for reporting. Suggestions?

I managed to catch the error in gdb, but the info is not that helpful:

Free Pascal Compiler version 3.1.1 [2016/08/23] for x86_64 Copyright (c) 1993-2016 by Florian Klaempfl and others (1002) Target OS: Linux for x86-64
(3104) Compiling pgopher.lpr

Breakpoint 3, RESOLVE (this=...) at symtype.pas:823 823 internalerror(200310221);
(gdb) where
#0 RESOLVE (this=...) at symtype.pas:823
#1 0x0217000100000008 in ?? ()
#2 0x00007fff9e000000 in ?? ()
#3 0x00007ffff7754f00 in ?? ()
#4 0x00007fffe391fac0 in ?? ()
#5 0x0000000000000000 in ?? ()
(gdb) info locals
result = 0x0
RESOLVE = 0x0
RESULT = 0x0
PM = 0x7fffe6e2f2c0
TYP = 228
IDX = 929
I = 140737035875040
LEN = 8
DATA = {1, 0, 23, 2, 0, 0, 0, 158, 255, 127, 0, 0, 0, 79, 117, 247, 255, 127, 0, 0, 192, 250, 145, 227, 255, 127, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 160, 0, 0, 0, 0, 0, 0, 0, 232, 227, 211, 226, 255, 127, 0, 0, 121, 135, 65, 0, 0, 0, 0, 0, 224, 55, 255, 255, 255, 127, 0, 0, 128, 0, 0, 0, 0, 0,
0, 0, 232, 227, 211, 226, 255, 127, 0, 0, 216, 25, 113, 231, 255, 127, 0, 0, 0, 26, 0, 0, 0, 0, 0, 0, 254, 126, 65, 0, 0, 0, 0, 0, 224, 26, 0, 0,
0, 0, 0, 0, 236, 140, 65, 0, 0, 0, 0, 0, 1, 180, 206, 226, 255, 127, 0, 0, 249, 137, 65, 0, 0, 0, 0, 0, 48, 79, 117, 247, 255, 127, 0, 0, 64, 82,
175, 240, 255, 127, 0, 0, 86, 3, 0, 0, 0, 0, 0, 0, 216, 33, 0, 0, 0, 0, 0, 0, 104, 36, 113, 247, 255, 127, 0, 0, 129, 141, 65, 0, 0, 0, 0, 0, 86,
3, 0, 0, 0, 0, 0, 0, 59, 4, 0, 0...}
(gdb) list
818 if dataidx<0 then
819 internalerror(200306067);
820 { read data }
821 current_module.derefdata.seek(dataidx); 822 if current_module.derefdata.read(len,1)1 then 823 internalerror(200310221);
824 if len>0 then
825 begin
826 if current_module.derefdata.read(data,len)len then 827 internalerror(200310222);

Colin

a backtrace ("bt" or "bt full") would probably have been better...

if the program is threaded, then
   thread apply all bt full

would have given additional info that might have been useful...

Not really. I've also started encountering it sometimes while working on the compiler, but haven't had the time yet to look for the cause.

Jonas

In fact it provides the same information in this case: (gdb) thread apply all bt full

Thread 1 (process 5541):
#0 INTERNALERROR (I=200310221) at verbose.pas:572 No locals.
#1 0x000000000048b81b in RESOLVE (this=<error reading variable: Cannot access memory at address 0xbf07dcd>) at symtype.pas:823 result = 0x0
RESOLVE = 0x0
RESULT = 0x0
PM = 0x7fffe6e2f2c0
TYP = 59
IDX = 925
I = 200310221
LEN = 8
DATA = {1, 0, 23, 2, 0, 0, 0, 158, 255, 127, 0, 0, 0, 79, 117, 247, 255, 127, 0, 0, 0, 255, 149, 227, 255, 127, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 160, 0, 0, 0, 0, 0, 0, 0, 136, 116, 211, 226, 255, 127, 0, 0, 121, 135, 65, 0, 0, 0, 0, 0, 224, 55, 255, 255, 255, 127, 0, 0, 128, 0, 0, 0, 0, 0, 0, 0, 136, 116, 211, 226, 255, 127, 0, 0, 56, 160, 162, 233, 255, 127, 0, 0, 0, 26, 0, 0, 0, 0, 0, 0, 254, 126, 65, 0, 0, 0,
0, 0, 224, 26, 0, 0, 0, 0, 0, 0, 236, 140, 65, 0, 0, 0, 0, 0, 1, 189, 206, 226, 255, 127, 0, 0, 249, 137, 65, 0, 0, 0, 0, 0, 48, 79, 117, 247, 255, 127, 0, 0, 192, 84, 175, 240, 255, 127, 0, 0, 86, 3, 0, 0, 0, 0, 0, 0, 216, 33, 0, 0, 0, 0, 0, 0, 104, 36, 113, 247, 255, 127,
0, 0, 129, 141, 65, 0, 0, 0, 0, 0, 86, 3, 0, 0, 0, 0, 0, 0, 59, 4, 0, 0...}
#2 0x0217000100000008 in ?? ()
No symbol table info available.
#3 0x00007fff9e000000 in ?? ()
No symbol table info available.
#4 0x00007ffff7754f00 in ?? ()
No symbol table info available.
#5 0x00007fffe395ff00 in ?? ()
No symbol table info available.
#6 0x0000000000000000 in ?? ()
No symbol table info available.