arm64: gdb(4) stub
ClosedPublic
Actions

Authored by mhorne on Dec 22 2020, 6:00 PM.

Details

Reviewers

manu
andrew

Group Reviewers

arm64

Commits

rGbbfa199cbc16: arm64: gdb(4) machine-dependent bits

Summary

This adds the basic support required for debugging with gdb over remote serial protocol.

Includes register read/write for general purpose and status registers, but not floating point registers at this time.

The GDB register numbers are defined here: https://sourceware.org/git/?p=binutils-gdb.git;a=blob;f=gdb/arch/aarch64.h;h=b75352461a412e74aa8e397507ff76a1824de1c7;hb=HEAD

Test Plan

Tested most common usage patterns, including: backtraces, listing/switching threads, setting breakpoints, inspecting/modifying kernel memory.

Diff Detail

Repository

rG FreeBSD src repository

Lint

Lint Not Applicable

Unit

Tests Not Applicable

Event Timeline

mhorne created this revision.Dec 22 2020, 6:00 PM

Herald added a reviewer: manu. · View Herald TranscriptDec 22 2020, 6:00 PM

Herald added subscribers: Contributor Reviews (src), emaste, andrew, imp. · View Herald Transcript

mhorne requested review of this revision.Dec 22 2020, 6:00 PM

Harbormaster completed remote builds in B35624: Diff 81066.Dec 22 2020, 6:01 PM

mhorne retitled this revision from Summary: to arm64: gdb(4) stub.Dec 22 2020, 6:06 PM

mhorne edited the summary of this revision. (Show Details)

mhorne edited the test plan for this revision. (Show Details)

mhorne added reviewers: andrew, arm64.

mhorne added parent revisions: D27720: arm64: remove pcb_pc, D27719: arm64: don't pass user trapframe to kdb_trap().

mhorne removed a subscriber: Contributor Reviews (src).

I will be on holiday for the next couple weeks, but posting this now in case anyone wants to look it over during that time. No rush to review.

mhorne edited the summary of this revision. (Show Details)Dec 22 2020, 6:09 PM

Adding @stevek as he is interested in this area.

An FYI rather than a review: I'm using these patches successfully on my RPi4, and it's proving invaluable. I'm eager to see these committed back to the main tree as soon as everyone feels ready.

emaste added inline comments.Jan 6 2021, 3:01 PM

sys/conf/files.arm64
171	We ought to move this to generic sys/conf/files

emaste added inline comments.Jan 6 2021, 3:15 PM

sys/arm64/arm64/gdb_machdep.c
78	Maybe worth a comment about updating both kdb_frame and the thrctx?

mhorne added inline comments.Jan 6 2021, 3:18 PM

sys/conf/files.arm64
171	How do you mean? Currently it is no different than other machdep files and should only be included here.

emaste added inline comments.Jan 6 2021, 3:24 PM

sys/conf/files.arm64
171	Oh yes, ignore that comment :)

andrew added inline comments.Jan 7 2021, 1:46 PM

sys/arm64/arm64/gdb_machdep.c
54–57	I find some of these hard to follow when they are all on one line.
62	Does this work in gdb or does it show duplicate stack frames?
sys/arm64/include/gdb_machdep.h
78	Duplicate
82	Duplicate

mhorne added inline comments.Jan 7 2021, 3:59 PM

sys/arm64/arm64/gdb_machdep.c
54–57	I agree, this style is a little odd but is used in other gdb_machdep implementations. I'm not attached to it however. Do you find it any better if they were aligned like this?
62	It does work in gdb without duplicates. I believe the reason is that `cpu_switch()`, where the PC actually is for sleeping threads, doesn't construct a new stack frame. For stopped threads, the same is true of `savectx()`, called in `ipi_stop()`.