Kyle Huey khuey

khuey commented on issue mozilla/rr#1451
@khuey

Where does the "unknown syscall" bit mentioned in the summary come from? If the that is what is going on we need the syscallno.

khuey commented on issue raspberrypi/linux#766
khuey commented on issue raspberrypi/linux#766
@khuey

Yeah, that's what it is. gdb has logic to pick the EABI out of the binary, that runs only if the OS/ABI is ELFOSABI_NONE. https://sourceware.org/gi…

khuey commented on issue raspberrypi/linux#766
@khuey

@nbrachet so after playing around for a while I noticed that if OS/ABI (in the readelf -h output) says "UNIX - System V" gdb works, but if it says …

khuey commented on issue raspberrypi/linux#766
@khuey

@nbrachet Good question! If you strace a gdb session do you see a PTRACE_POKEDATA/PTRACE_POKETEXT call with 0xef9f0001 as an argument? What is the …

khuey commented on issue raspberrypi/linux#766
@khuey

Unfortunately I don't know. It's not needed to run gdb, certainly.

khuey pushed to arm at khuey/rr
@khuey
khuey commented on issue raspberrypi/linux#766
@khuey

I didn't check if this is what is tripping gdb up, but despite the name, readelf says that ld-2.13.so is not built with hard float enabled.

khuey commented on issue raspberrypi/linux#766
@khuey

This is caused by gdb attempting to use an OABI syscall instruction to invoke the __ARM_NR_breakpoint syscall. The Raspberry Pi kernel is built wit…

khuey commented on issue mozilla/rr#1433
@khuey

So, FWIW, AWS and GCP do not have perf counters enabled, and Azure has AMD hardware.

khuey commented on pull request mozilla/rr#1438
@khuey

Aha, I had missed that there are separate definitions of __kernel_time_t. Alright, yes, we'll need to split it up by arch.

khuey commented on pull request mozilla/rr#1438
@khuey

I don't understand. The issue here is structure packing. The sizes of the variables themselves are already correct on all three architectures, but …

khuey commented on pull request mozilla/rr#1438
@khuey

Yes, that's what it "breaks". On Mar 11, 2015 7:12 PM, "Nathan Froyd" notifications@github.com wrote: Huh, weird. If it was structure packing, I'd…

khuey commented on pull request mozilla/rr#1438
@khuey

I'm not entirely sure. I imagine its a structure packing issue.

khuey opened pull request mozilla/rr#1438
@khuey
Copy the kernel's definition of msgid64_ds verbatim.
1 commit with 12 additions and 3 deletions
khuey created branch msgid64_ds at khuey/rr
khuey pushed to master at khuey/rr
@khuey
  • @rocallahan 01c4207
    Add watchpoint_syscall to test firing of watchpoints on syscalls that…
  • @rocallahan 9c76c94
    Use AddressSpace watchpoint monitoring to check for actual watchpoint…
  • 85 more commits »
khuey pushed to arm at khuey/rr
@khuey
khuey pushed to dev at khuey/gecko-dev
@khuey dev is now 2a7843e
khuey pushed to dev at khuey/gecko-dev
@khuey
khuey commented on issue mozilla/rr#1112
@khuey

Long term we need to figure out if cmake can sanely support cross compiling. I wasn't able to find much about doing that online last year, unfortun…

khuey commented on issue mozilla/rr#1424
@khuey

Ordering of the different threads varies a lot from run to run, so it's not easy to follow. I can try figuring it out later this week.

khuey commented on issue mozilla/rr#1424
@khuey

Unfortunately there isn't much of a stack. The thread appears to be in a wait4. How can I figure out which executable the thread corresponds to to …

khuey commented on issue mozilla/rr#1424
@khuey

After reading #1421 I do basically have the same issue. The problem is that attaching the emergency debugger via "target remote :pid" tries to perf…

khuey pushed to dev3 at khuey/gecko-dev
@khuey
khuey commented on issue mozilla/rr#1424
@khuey

Yes.

khuey commented on issue mozilla/rr#1424
@khuey

It's 100% reproducible, at least on my machine. Trying to emergency debug it causes rr to exit though. gdb prints out warning: limiting remote sugg…

khuey commented on issue mozilla/rr#1424
@khuey

This happens just after a SIGCHLD. [Signal] 5273: handling signal SIGCHLD (pevent: 0, event: (none) [RecordSession] line 73813: Active task is 5273…

khuey opened issue mozilla/rr#1424
@khuey
rr fails to record mach help
@khuey