On Thu, Mar 03, 2011 at 11:14:59AM +0000, Robert Watson wrote: > > On Tue, 1 Mar 2011, Dmitry Chagin wrote: > > > Teach kdump to decode linux syscalls names too. > > > > Fix bug introduced in my previous commit: the kernel always dump native > > signal numbers, so no need to check the ABI in ktrpsig(). > > Does this mean that we're eliminating the need for the long-broken > linux_kdump? >
mostly yes, w/o syscall parameter decoding. > Is there any ABI record in ktrace files, and/or should we add one somehow? > Now kernel writes sv_flags to the trace file at fork, exec, exit time. And kdump knows how to handle this. > I'd love to be able to process a kdump file generated from a blend of ABIs: > i.e., if I do ktrace -di tcsh, and then run a bunch of commands from > different > ABIs (FreeBSD/32-bit Intel, FreeBSD/64-bit Intel, Linux/32-bit Intel, and > possible FreeBSD/64 MIPS so that a distributed ktracing tool gathers the > results across hosts for a single action), have kdump Just Work. This > suggests that a per-record ABI indicator might be useful. > > Robert This is the next stage :) -- chd, Have fun! >
pgpEGh6n8rKm0.pgp
Description: PGP signature
