2002-04-04 21:04:09 +02:00
|
|
|
- what's the byte-order of [gs]et_{fp,}reg?
|
|
|
|
|
|
|
|
|
|
- locking
|
|
|
|
|
|
|
|
|
|
- memory allocation
|
|
|
|
|
|
2002-04-02 01:01:22 +02:00
|
|
|
- unify UNW_IA64_AR_BSP and UNW_IA64_BSP
|
|
|
|
|
|
2002-04-04 21:04:09 +02:00
|
|
|
- finish & test unwinding across signal frame (pay attention that signal
|
|
|
|
|
frame gets updated)
|
|
|
|
|
|
|
|
|
|
- API for dynamically generated code and redirection
|
|
|
|
|
|
2002-07-18 05:58:34 +02:00
|
|
|
- to test:
|
|
|
|
|
o verify correctness of spill-area layout
|
|
|
|
|
o verify working of frgr_mem descriptor (especially for f16-f31)
|
|
|
|
|
|
2002-04-06 01:37:55 +02:00
|
|
|
+ multi-architecture support?
|
2002-04-04 21:04:09 +02:00
|
|
|
+ debuggers need additional frame info:
|
2002-04-02 01:01:22 +02:00
|
|
|
- ability to determine first non-prologue instruction
|
|
|
|
|
(gdb calls this skip_prologue) => this doesn't really
|
|
|
|
|
make much sense for optimized code: there may be now
|
|
|
|
|
single IP at which the .body starts; the gdb maintainers
|
|
|
|
|
are aware of this and are planning to modify gdb such that
|
|
|
|
|
skip_prologue can always return 0
|
|
|
|
|
- ability to find out whether and where a frame saved a register
|
