2.6.23-rc6-mm1: kgdb support on ppc64 utterly broken

On (18/09/07 18:44), Mel Gorman didst pronounce:

Hi Andrew,

IA64 failed to build in kgdb with the messages

ppc64 kgdb support is also broken but in a much more fundamental manner.
allmodconfig shows up

In file included from include/linux/kgdb.h:22,
                 from arch/powerpc/kernel/legacy_serial.c:15:
include/asm/kgdb.h:34: error: 'debugger' redeclared as different kind of symbol
include/asm/system.h:85: error: previous definition of 'debugger' was here
include/asm/kgdb.h:35: error: 'debugger_bpt' redeclared as different kind of symbol
include/asm/system.h:87: error: previous definition of 'debugger_bpt' was here
include/asm/kgdb.h:36: error: 'debugger_sstep' redeclared as different kind of symbol
include/asm/system.h:88: error: previous definition of 'debugger_sstep' was here
include/asm/kgdb.h:37: error: 'debugger_iabr_match' redeclared as different kind of symbol
include/asm/system.h:89: error: previous definition of 'debugger_iabr_match' was here
include/asm/kgdb.h:38: error: 'debugger_dabr_match' redeclared as different kind of symbol
include/asm/system.h:90: error: previous definition of 'debugger_dabr_match' was here
include/asm/kgdb.h:39: error: 'debugger_fault_handler' redeclared as different kind of symbol
include/asm/system.h:91: error: previous definition of 'debugger_fault_handler' was here
In file included from arch/powerpc/kernel/legacy_serial.c:15:
include/linux/kgdb.h:69: error: `BREAK_INSTR_SIZE' undeclared here (not in a function)
make[1]: *** [arch/powerpc/kernel/legacy_serial.o] Error 1
make: *** [arch/powerpc/kernel] Error 2

I cleared the redeclarations up and put in some defines but there is
much more fundamental breakage like

kernel/kgdb.c:122: error: `BUFMAX' undeclared here (not in a function)
kernel/kgdb.c:128: error: `NUMCRITREGBYTES' undeclared here (not in a function)
kernel/kgdb.c: In function `write_mem_msg':
kernel/kgdb.c:554: error: `CACHE_FLUSH_IS_SAFE' undeclared (first use in this function)
kernel/kgdb.c:554: error: (Each undeclared identifier is reported only once
kernel/kgdb.c:554: error: for each function it appears in.)
kernel/kgdb.c: In function `getthread':
kernel/kgdb.c:615: error: implicit declaration of function `kgdb_get_shadow_thread'
kernel/kgdb.c:616: warning: return makes pointer from integer without a cast
kernel/kgdb.c: In function `kgdb_activate_sw_breakpoints':
kernel/kgdb.c:719: error: `CACHE_FLUSH_IS_SAFE' undeclared (first use in this function)
kernel/kgdb.c: In function `kgdb_deactivate_sw_breakpoints':
kernel/kgdb.c:784: error: `CACHE_FLUSH_IS_SAFE' undeclared (first use in this function)
kernel/kgdb.c: In function `kgdb_handle_exception':
kernel/kgdb.c:956: error: implicit declaration of function `kgdb_skipexception'
kernel/kgdb.c:1036: error: implicit declaration of function `kgdb_disable_hw_debug'
kernel/kgdb.c:1072: error: implicit declaration of function `kgdb_post_master_code'
kernel/kgdb.c:1153: error: implicit declaration of function `kgdb_shadow_regs'
kernel/kgdb.c:1156: warning: assignment makes pointer from integer without a cast
kernel/kgdb.c:1333: error: implicit declaration of function `kgdb_shadowinfo'
kernel/kgdb.c: In function `breakpoint':
kernel/kgdb.c:1891: error: implicit declaration of function `BREAKPOINT'
kernel/kgdb.c: At top level:
kernel/kgdb.c:122: error: storage size of `remcom_in_buffer' isn't known
kernel/kgdb.c:123: error: storage size of `remcom_out_buffer' isn't known
kernel/kgdb.c:129: error: storage size of `kgdb_fault_jmp_regs' isn't known
kernel/kgdb.c:853: error: storage size of `gdbmsgbuf' isn't known
kernel/kgdb.c:122: warning: 'remcom_in_buffer' defined but not used
kernel/kgdb.c:123: warning: 'remcom_out_buffer' defined but not used
kernel/kgdb.c:853: warning: 'gdbmsgbuf' defined but not used
make[1]: *** [kernel/kgdb.o] Error 1
make: *** [kernel] Error 2

Even with the defines fixed up, I don't know what to do about
kgdb_skipexception() and friends because frankly I don't know what I'm
doing with kgdb. linuxppc-dev added to cc list this time in case they have
something useful.

The best I was able to come up with as a candidate fix was to disable
kgdb on ppc64 altogether. This is a cop-out, not a fix, sorry.

Signed-off-by: Mel Gorman <redacted>
---
 Kconfig.kgdb |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff -rup -X /usr/src/patchset-0.6/bin//dontdiff linux-2.6.23-rc6-mm1-020_fix_acpi/lib/Kconfig.kgdb linux-2.6.23-rc6-mm1-025_fix_ppc64_kgdb/lib/Kconfig.kgdb

--- linux-2.6.23-rc6-mm1-020_fix_acpi/lib/Kconfig.kgdb	2007-09-18 11:29:30.000000000 +0100
+++ linux-2.6.23-rc6-mm1-025_fix_ppc64_kgdb/lib/Kconfig.kgdb	2007-09-19 15:54:23.000000000 +0100

@@ -14,7 +14,7 @@ config KGDB
 	bool "KGDB: kernel debugging with remote gdb"
 	select WANT_EXTRA_DEBUG_INFORMATION
 	select KGDB_ARCH_HAS_SHADOW_INFO if X86_64
-	depends on DEBUG_KERNEL && (ARM || X86 || MIPS || (SUPERH && !SUPERH64) || IA64 || PPC)
+	depends on DEBUG_KERNEL && (ARM || X86 || MIPS || (SUPERH && !SUPERH64) || IA64)
 	help
 	  If you say Y here, it will be possible to remotely debug the
 	  kernel using gdb.  Documentation of kernel debugger is available

-- 
Mel Gorman
Part-time Phd Student                          Linux Technology Center
University of Limerick                         IBM Dublin Software Lab