On Thu 23 Jul 2009 00:02, gyang pondered: > On Wed, 2009-07-22 at 09:47 -0400, Robin Getz wrote: > > > > Don't you want the Level3 SEQSTAT on the frame? otherwise - it will > > print out the wrong seqstat when the traps_c code is run. > > > I now find that the SEQSTAT is not a problem. But > cpu_pda[].dcplb_fault_addr will be wrong in trap_c.
OK - that just means the location of where we are storing the cpu_pda[].dcplb_fault_addr is wrong... We store them in _ex_trap_c() when CONFIG_DEBUG_DOUBLEFAULT is not set, and _trap when CONFIG_DEBUG_DOUBLEFAULT is set... Hmmm.... This is the problem... In the normal situation (CONFIG_DEBUG_DOUBLEFAULT is off) we don't care about normal DCPLB/ICPLB misses - and shouldn't keep track of this. When we are debugging an doublefault - we need to keep track of everything - since it might be the DCPLB/ICPLB miss which is causing the error... I guess we need to add a [id]cplb_doublefault_addr - and have that dependant on CONFIG_DEBUG_DOUBLEFAULT? Maybe? _______________________________________________ Linux-kernel-commits mailing list [email protected] https://blackfin.uclinux.org/mailman/listinfo/linux-kernel-commits
