Use the context created in makectx() for stack traces.

[jhb]

Summary

Always use the kdb_thr_ctx() for db_trace_thread() as on other
architectures. Initialize pcb_ra to be the sepc from the saved
trapframe rather than the saved ra to avoid skipping a frame.

Test Plan

• compared traces from ddb

[jhb]

Before:

KDB: enter: sysctl debug.kdb.enter
[ thread pid 690 tid 100049 ]
Stopped at      kdb_trap+0x42a: sd      zero,0(a0)
db> tr
Tracing pid 690 tid 100049 td 0xffffffd001a26560
db_trace_self() at db_trace_self
db_hex2dec() at db_hex2dec+0x1d2
db_command_loop() at db_command_loop+0x2c6
db_command_loop() at db_command_loop+0x6a
db_fetch_ksymtab() at db_fetch_ksymtab+0x2c8
kdb_trap() at kdb_trap+0x156
do_trap_supervisor() at do_trap_supervisor+0xaa
cpu_exception_handler_supervisor() at cpu_exception_handler_supervisor+0x6a
--- exception 3, tval = 0
kdb_trap() at kdb_trap+0x42a
sbuf_new_for_sysctl() at sbuf_new_for_sysctl+0x11a
kernel_sysctl() at kernel_sysctl+0x376
userland_sysctl() at userland_sysctl+0x150
sys___sysctl() at sys___sysctl+0x54
do_trap_user() at do_trap_user+0x3d6
cpu_exception_handler_user() at cpu_exception_handler_user+0x74
--- exception 8, tval = 0
db> c

vs after:

KDB: enter: sysctl debug.kdb.enter
[ thread pid 700 tid 100063 ]
Stopped at      kdb_trap+0x42a: sd      zero,0(a0)
db> tr
Tracing pid 700 tid 100063 td 0xffffffd03e80c560
kdb_trap() at kdb_trap+0x428
sbuf_new_for_sysctl() at sbuf_new_for_sysctl+0x11a
kernel_sysctl() at kernel_sysctl+0x376
userland_sysctl() at userland_sysctl+0x150
sys___sysctl() at sys___sysctl+0x54
do_trap_user() at do_trap_user+0x3d6
cpu_exception_handler_user() at cpu_exception_handler_user+0x74
--- exception 8, tval = 0
db> c

Mostly it means stack traces don't show the internals of being in ddb itself but start at the time of entering the debugger.

[mhorne]

LGTM. The change to db_trace_thread also needs doing for arm and arm64.

[jhb]

Yes, it was probably copied from arm64 originally. Fixing it on those platforms requires auditing makectx() to make sure it DTRT however.

[mhorne]

In D23513#516255, @jhb wrote:

Yes, it was probably copied from arm64 originally. Fixing it on those platforms requires auditing makectx() to make sure it DTRT however.

Ack, thanks.