Splitting and resending for gdb-patches since the limit is 400k for it. On Tue, 12 Jun 2012 16:34:32 +0200, Jan wrote: > On Tue, 12 Jun 2012 11:55:43 +0200, Jan Kratochvil wrote: > > On Mon, 11 Jun 2012 20:33:36 +0200, Siddhesh Poyarekar wrote: > > > splint +posixlib +gnuextensions -linelen 999 -hints > > > -showcolumn \ -Isplint -exportlocal -DTUI -I. \ > > > -Icommon -I../include -I../bfd -I../libdecnumber > > > -I../opcodes \ -I.. -I/usr/include/python2.7 $i &>splint/$i.out > [...] > > But apparently maybe just removing some assertion would be enough > > for this GDB purpose. > > Maybe for example removing +posixlib would be easier. I have attached an updated patch (and changelog) after a minor rebase and minor changes you suggested earlier (lines longer than 80 chars, pulongest for size_t). New command is now: for i in `cat files`; do mkdir -p splint/`dirname $i` splint -linelen 999 -hints -showcolumn +trytorecover \ -Dsigjmp_buf=int -Dsiginfo_t=int \ -Isplint -exportlocal -DTUI -I. \ -Icommon -I../include -I../bfd -I../libdecnumber \ -I../opcodes -I.. -I/usr/include/python2.7 \ $i &>splint/$i.out done The command now fails only for the following arch dependent files (no additional assertion failures): ia64-linux-nat.c mips-linux-nat.c nto-procfs.c procfs.c s390-nat.c on parser errors due to missing headers. Attached new locdiff output as well as modified my script for the following: * Skip 'Test expression for ... not boolean' warnings * Skip 'New fresh storage ... passed as implicitly temp' warnings since our changes couldn't have caused them. * Skip 'Operands have incompatible types' when the types after the change are ULONGEST and LONGEST * Remove blocks that do not record changes, i.e. of the type: --- 8< ----- LOC... | foo... |=> Remove this. bar... | --- 8< ----- [ -+]LOC... If these changes are OK, then I'll record my review on the resultant locdiff.processed.out, which is 2330 lines now. Thanks, Siddhesh