I finally managed to catch one of these errors on amd64 machine. The issue is that on some packages that are expected to fail, the failure results on an exception.However and despite the expected failure is caught, the test infrastructure detects a message written to stderr and thus triggers the failure. I've rebuilt lazuild using debug symbols, and the error seems to be caught in an object destructor (#3) as shown in the following gdb stack trace:Warning: (lazarus) [TLazPackageGraph.ConvertPackageRSTFiles] skipping read only directory /usr/lib/lazarus/2.0.6/components/sparta/dockedformeditor/language/ Thread 4 "lazbuild" received signal SIGSEGV, Segmentation fault.[Switching to Thread 0x7ffff7408700 (LWP 25224)]0x000000000051ac03 in LEAVECRITICALSECTION (this=0x7ffff7649720) at ideexterntoolintf.pas:12371237 FWorkerMessages.LeaveC riticalSection;(gdb) bt#0 0x000000000051ac03 in LEAVECRITICALSECTION (this=0x7ffff7649720) at ideexterntoolintf.pas:1237#1 0x00000000009e9cb5 in SETTHREAD (this=0x7ffff7649720, AVALUE=0x0) at exttools.pas:411#2 0x00000000009edc57 in SETTOOL (this=0x7ffff761fc40, AVALUE=0x0) at exttools.pas:1313#3 0x00000000009eeeb1 in DESTROY (this=0x7ffff761fc40, vmt=0x1) at exttools.pas:1612#4 0x000000000041cd20 in SYSTEM$_$TOBJECT_$__$$_FREE ()#5 0x00007ffff77a9a80 in ?? ()#6 0x0000000000440cd7 in CLASSES_$$_THREADFUNC$POINTER$$INT64 ()#7 0x00007ffff761fc40 in ?? ()#8 0x0000000000000000 in ?? ()(gdb) To be continued! -- Cheers, Abou Al Montacir
signature.asc
Description: This is a digitally signed message part
