Re: [Crash-utility] Re: [RFC]Patch to accomodate cpu_pda changesin x86_64 kernels

On Wed, Feb 01, 2006 at 08:47:07AM -0500, Dave Anderson wrote: > > That should do it. > > Minor nits: for clarity's sake, I'd make the local variable have the > same name as the kernel symbol it's representing (i.e. _cpu_data > instead of __cpu_data), just check it as a boolean instead of it > being == 1, and increment cpu_pda by sizeof(void *). Hi Dave Incorporating the suggested changes and resending the patch. Kindly review. Thanks Rachita Signed-off-by: Rachita Kothiyal <rachita(a)in.ibm.com&gt; --- defs.h | 9 +++++++ x86_64.c | 77 +++++++++++++++++++++++++++++++++++++++++++++++---------------- 2 files changed, 67 insertions(+), 19 deletions(-) diff -puN x86_64.c~crash-fix-cpu-pda x86_64.c --- crash-4.0-2.19/x86_64.c~crash-fix-cpu-pda 2006-02-01 20:08:14.000000000 +0530 +++ crash-4.0-2.19-rachita/x86_64.c 2006-02-01 20:59:59.033113672 +0530 @@ -365,9 +365,9 @@ x86_64_dump_machdep_table(ulong arg) static void x86_64_cpu_pda_init(void) { - int i, cpus, nr_pda, cpunumber; + int i, cpus, nr_pda, cpunumber, _cpu_pda; char *cpu_pda_buf; - ulong level4_pgt, data_offset; + ulong level4_pgt, data_offset, cpu_pda_addr; struct syment *sp, *nsp; ulong offset, istacksize; @@ -383,12 +383,26 @@ x86_64_cpu_pda_init(void) cpu_pda_buf = GETBUF(SIZE(x8664_pda)); - if (!(nr_pda = get_array_length("cpu_pda", NULL, 0))) - nr_pda = NR_CPUS; + if (symbol_exists("_cpu_pda")) { + if (!(nr_pda = get_array_length("_cpu_pda", NULL, 0))) + nr_pda = NR_CPUS; + _cpu_pda = TRUE; + } else { + if (!(nr_pda = get_array_length("cpu_pda", NULL, 0))) + nr_pda = NR_CPUS; + _cpu_pda = FALSE; + } for (i = cpus = 0; i < nr_pda; i++) { - if (!CPU_PDA_READ(i, cpu_pda_buf)) - break; + if (_cpu_pda) { + cpu_pda_addr = 0; + if (!_CPU_PDA_READ(i, cpu_pda_buf)) + break;