Opened 14 years ago
Closed 14 years ago
#6160 closed enhancement (fixed)
[PATCH] ppc: Implement arch_debug_save_registers and shorten stack traces
| Reported by: | andreasf | Owned by: | andreasf |
|---|---|---|---|
| Priority: | normal | Milestone: | R1 |
| Component: | System/Kernel | Version: | R1/Development |
| Keywords: | Cc: | ||
| Blocked By: | Blocking: | ||
| Platform: | PowerPC |
Description
As suggested in #6139, here's a patch saving the stack pointer in arch_debug_save_registers and retrieving it for the stack trace instead of using the current function's stack frame.
This was tested to exclude panic and upwards from both the immediate KDL stack trace and the interactive bt stack trace.
Attachments (2)
Change History (7)
by , 14 years ago
| Attachment: | 0001-ppc-Save-debug-register-state-and-shorten-stack-trac.patch added |
|---|
comment:1 by , 14 years ago
| patch: | 0 → 1 |
|---|
comment:2 by , 14 years ago
| Owner: | changed from to |
|---|---|
| Status: | new → in-progress |
comment:3 by , 14 years ago
| patch: | 1 → 0 |
|---|
by , 14 years ago
| Attachment: | 0001-ppc-Drop-one-frame-in-stack-trace.patch added |
|---|
draft patch: drop "panic" from stack trace
comment:4 by , 14 years ago
| patch: | 0 → 1 |
|---|
comment:5 by , 14 years ago
| Resolution: | → fixed |
|---|---|
| Status: | in-progress → closed |
I've applied slightly different patches in hrev37281 and hrev37282.
In particular I've fixed a type inconsistency (uint32 vs. addr_t) and save the proper r1 value rather than the one that's nice to have for the stack trace.
I've attached a potential add-on patch that does the additional stack frame dereference on the stack_trace() side of things, but I believe this behavior should be synchronized with x86. Before we reach a point with working USB drivers we cannot test the alternative (non-panic) KDL entry through keyboard shortcut anyway. Closing this ticket for now.
proposed patch