Re: Bug in powerpc gdb?
Roger. I suspect that some argument variables not being available is the result of optimization or normal program flow. Roger Leigh wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Till Straumann <[EMAIL PROTECTED]> writes: -- snp (gdb) bt #0 sbuild_session_setup_chroot (session=0x1004e770, session_chroot=0x1004bf60, setup_type=SBUILD_CHROOT_SETUP_START, error=0x7fc04548) at sbuild-session.c:491 #1 0x10014488 in sbuild_session_run (session=0x1004e770, error=0x10054788) at sbuild-session.c:1064 #2 0x10015304 in sbuild_cclosure_marshal_BOOLEAN__BOXED (closure=0x1004e0e0, return_value=0x7fc04680, n_param_values=Variable "n_param_values" is not available. ) at sbuild-marshallers.c:162 If the register holding 'n_param_values' is overwritten/reused at the time g_type_class_meta_marhal() is called then there is (in general) no way for the unwinder to resurrect the value it held earlier. #3 0x0fa7d9dc in g_type_class_meta_marshal (closure=0x1004e770, return_value=0x7fade680, n_param_values=2, param_values=0x7fade790, invocation_hint=0x7fade6a0, marshal_data=0x48) at gclosure.c:569 I'm not sure why some variables are "not available", but at least it can now correctly walk through the stack. good :-) I submitted PR2029 to the GDB bug database -- please use the patch from there and not the earlier version I posted yesterday as it contains an error. --T. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Bug in powerpc gdb?
I also experienced problems with the stack unwinder when using more recent gcc optimization. The attached patch may help (I thought I had sent it to the GDB site but am unable to find it in the database). HTH -- Till (CC me for questions/comments, I'm not subscribed) Patch against gdb-6.2.1. To apply, run patch -p0 < 'this_file' from the gdb-6.2.1 topdir. It's always a good idea to try with the '--dry-run' option first. Provides: fixes a bug (triggered by newer gcc optimizations) in the rs6000/powerpc stack unwinder. Prerequisites: n/a Till Straumann, 3/14/2005 Index: gdb/rs6000-tdep.c === RCS file: /afs/slac/g/spear/cvsrep/tools/gdb/gdb/rs6000-tdep.c,v retrieving revision 1.1.1.1 retrieving revision 1.2 diff -c -r1.1.1.1 -r1.2 *** gdb/rs6000-tdep.c 10 Nov 2004 18:25:48 - 1.1.1.1 --- gdb/rs6000-tdep.c 10 Nov 2004 18:39:09 - 1.2 *** *** 1184,1190 the search if we have no valid frame yet or if the return address is not yet saved in the frame. */ if (fdata->frameless == 0 ! && (lr_reg == -1 || fdata->nosavedpc == 0)) break; if (op == 0x4e800020 /* blr */ --- 1184,1190 the search if we have no valid frame yet or if the return address is not yet saved in the frame. */ if (fdata->frameless == 0 ! && (lr_reg != -1 && fdata->nosavedpc == 0)) break; if (op == 0x4e800020 /* blr */
Re: PowerPC kernels, partitioning on macppc
Regarding PPC architecture, assembly etc.: The 32-bit PowerPC architecture and assembly language are extensively (including caches, MMU etc.) described e.g., here: http://www.freescale.com/files/product/doc/MPCFPE32B.pdf On the FreeScale (motorola) site you can also find documents describing AltiVec and processor specific (e.g., 'G4' vs. 'G3') implementation details. I recommend to also peeking at the SYSVR4 ABI specification which defines things like calling conventions, register usage and the like: http://www.omega-environment.com/documentation/SVR4abippc.pdf Note that as always these links may be obsolete shortly as the commercial world just loves to rearrange their websites faster than the weather changes. google is your friend. HTH -- Till PS: I'm not subscribed to the list. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
