No dump from x86 Android

[Antti Mattila]

I successfully get crash dumps with arm based devices, but running simple test app on x86 device does not generate crash dump. 

Registering crash handler like this:

        google_breakpad::MinidumpDescriptor descriptor(path);

        exceptionHandler = new google_breakpad::ExceptionHandler(descriptor, NULL, DumpCallback, NULL, false, -1);

but DumpCallback function is never called on x86 systems. Tried three different devices and simulator.  

Is this known issue, or does x86 require something special on Android?

[Antti Mattila]

To add couple of details. Api level is set to 15. Using latest NDK. 

This also happens with sample_app bundled with the project. Easy to replicate.

[Antti Mattila]

Also, when running this 

exceptionHandler->SimulateSignalDelivery(1);

with debugger, it seems to crash to sys_mmap call.

maanantai 27. maaliskuuta 2017 13.39.55 UTC+3 Antti Mattila kirjoitti:

[Jay Soffian]

Replying to a year-old message since I just came across this issue. 

LSS seems to not work properly when compiled for x86 with clang under Android. It segfaults on some calls. So far both sys_prctl and sys_mmap fail. Android NDK 16.1.4479499.

If I compile with gcc (NDK_TOOLCHAIN_VERSION := 4.9) it works as expected. I’m testing under the emulator, but it repros with a few different Android versions.

Current breakpad which is using this LSS revision:

https://chromium.googlesource.com/linux-syscall-support/+/e6527b0cd469e3ff5764785dadcb39bf7d787154

j.

--
You received this message because you are subscribed to the Google Groups "google-breakpad-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-discuss+unsub...@googlegroups.com.
To post to this group, send email to google-breakpad-discuss@googlegroups.com.
Visit this group at https://groups.google.com/group/google-breakpad-discuss.
For more options, visit https://groups.google.com/d/optout.

[Mike Frysinger]

is it actually segfaulting ?  or is it a different signal that is killing it ?

is there anything in the logs or kernel buffer ?

newer Android is using seccomp filters, so i'm wondering if it's tripping that.

-mike

On Tue, May 1, 2018 at 11:05 AM Jay Soffian <jayso...@gmail.com> wrote:

Replying to a year-old message since I just came across this issue. 

LSS seems to not work properly when compiled for x86 with clang under Android. It segfaults on some calls. So far both sys_prctl and sys_mmap fail. Android NDK 16.1.4479499.

If I compile with gcc (NDK_TOOLCHAIN_VERSION := 4.9) it works as expected. I’m testing under the emulator, but it repros with a few different Android versions.

Current breakpad which is using this LSS revision:

https://chromium.googlesource.com/linux-syscall-support/+/e6527b0cd469e3ff5764785dadcb39bf7d787154

j.

On Tue, Mar 28, 2017 at 6:18 AM, Antti Mattila <antti....@gmail.com> wrote:

Also, when running this 

exceptionHandler->SimulateSignalDelivery(1);

with debugger, it seems to crash to sys_mmap call.

maanantai 27. maaliskuuta 2017 13.39.55 UTC+3 Antti Mattila kirjoitti:

I successfully get crash dumps with arm based devices, but running simple test app on x86 device does not generate crash dump. 

Registering crash handler like this:

        google_breakpad::MinidumpDescriptor descriptor(path);

        exceptionHandler = new google_breakpad::ExceptionHandler(descriptor, NULL, DumpCallback, NULL, false, -1);

but DumpCallback function is never called on x86 systems. Tried three different devices and simulator.  

Is this known issue, or does x86 require something special on Android?

--
You received this message because you are subscribed to the Google Groups "google-breakpad-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-d...@googlegroups.com.
To post to this group, send email to google-brea...@googlegroups.com.

Visit this group at https://groups.google.com/group/google-breakpad-discuss.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "google-breakpad-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-d...@googlegroups.com.
To post to this group, send email to google-brea...@googlegroups.com.

[Jay Soffian]

It’s actually segfaulting, yes, at least according to logcat. I don’t think it’s seccomp because I’m testing with emulators based on API 19 and API 21. Also, the issue occurs when compiled with clang, but not with gcc, so I’m not sure how seccomp would come into play with one of those but not the other. Also, the same code works on arm/arm64 devices at various API levels, including the API levels where it’s failing on x86.

I’’ve tried both abort() and SimulateSignalDelivery() to trigger the handler. logcat -v time -b all -s ‘*:v’ looks like:

05-01 14:51:42.464 W/google-breakpad( 9938): PR_SET_DUMPABLE

05-01 14:51:42.464 W/AudioTrack( 1501): AUDIO_OUTPUT_FLAG_FAST denied by client

05-01 14:51:42.473 I/WindowState( 1501): WIN DEATH: Window{d1e2787 u0 …demo.DemoActivity}

05-01 14:51:42.474 W/libprocessgroup( 1501): failed to open /acct/uid_10083/pid_9938/cgroup.procs: No such file or directory

05-01 14:51:42.474 I/ActivityManager( 1501): Process …demo (pid 9938) has died

05-01 14:51:42.476 W/ActivityManager( 1501): Force removing ActivityRecord{cafb3b5 u0 …demo/.DemoActivity t37}: app died, no saved state

05-01 14:51:42.477 I/Zygote  ( 1152): Process 9938 exited due to signal (11)

I added a logging call just before sys_prctl(PR_SET_DUMPABLE, 1, 0, 0, 0) is called inside ExceptionHandler::HandleSignal. That’s the PR_SET_DUMPABLE line. The logging call immediately after it does not appear, so the segfault occurs during the sys_prctl call. If I switch that line to call prctl directly, then the segfault occurs when sys_mmap is called by the PageAllocator.

That’s on a Nexus_5X_API_21 avd. 

j.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-discuss+unsub...@googlegroups.com.
To post to this group, send email to google-breakpad-discuss@googlegroups.com.

Visit this group at https://groups.google.com/group/google-breakpad-discuss.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "google-breakpad-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-discuss+unsub...@googlegroups.com.
To post to this group, send email to google-breakpad-discuss@googlegroups.com.

Visit this group at https://groups.google.com/group/google-breakpad-discuss.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "google-breakpad-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-discuss+unsub...@googlegroups.com.
To post to this group, send email to google-breakpad-discuss@googlegroups.com.

[Jay Soffian]

Don’t know if this helps any but when I trigger the handler via SimulateSignalDeliver(SIGABRT) and with lldb attached:

- sys_prctl crashes with a fault address of 0x1c.

- sys_mmap crashes with a fault address of 0x3e.

j.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-discuss+unsubscr...@googlegroups.com.

To post to this group, send email to google-breakpad-discuss@googlegroups.com.
Visit this group at https://groups.google.com/group/google-breakpad-discuss.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "google-breakpad-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-discuss+unsubscr...@googlegroups.com.

To post to this group, send email to google-breakpad-discuss@googlegroups.com.
Visit this group at https://groups.google.com/group/google-breakpad-discuss.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "google-breakpad-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-discuss+unsubscr...@googlegroups.com.

[Jay Soffian]

More info. Learning as I go here ( I haven’t written assembly in a couple decades and it was 68000 then…). This is as compiled by clang.. 

(lldb) dis

libcrashmanager.so`sys_prctl:

    0xa65e3db0 <+0>:   pushl  %ebp

    0xa65e3db1 <+1>:   movl   %esp, %ebp

    0xa65e3db3 <+3>:   pushl  %ebx

    0xa65e3db4 <+4>:   pushl  %edi

    0xa65e3db5 <+5>:   pushl  %esi

    0xa65e3db6 <+6>:   andl   $-0x10, %esp

    0xa65e3db9 <+9>:   subl   $0x40, %esp

    0xa65e3dbc <+12>:  movl   %esp, %esi

    0xa65e3dbe <+14>:  calll  0xa65e3dc3                ; <+19> at linux_syscall_support.h:3479

    0xa65e3dc3 <+19>:  popl   %eax

    0xa65e3dc4 <+20>:  addl   $0x8c109, %eax            ; imm = 0x8C109 

    0xa65e3dca <+26>:  movl   0x18(%ebp), %ecx

    0xa65e3dcd <+29>:  movl   0x14(%ebp), %edx

    0xa65e3dd0 <+32>:  movl   0x10(%ebp), %edi

    0xa65e3dd3 <+35>:  movl   0xc(%ebp), %ebx

    0xa65e3dd6 <+38>:  movl   %eax, 0x34(%esi)

    0xa65e3dd9 <+41>:  movl   0x8(%ebp), %eax

    0xa65e3ddc <+44>:  movl   %eax, 0x30(%esi)

    0xa65e3ddf <+47>:  movl   0x8(%ebp), %eax

    0xa65e3de2 <+50>:  movl   %eax, 0x2c(%esi)

    0xa65e3de5 <+53>:  movl   0xc(%ebp), %eax

    0xa65e3de8 <+56>:  movl   %eax, 0x28(%esi)

    0xa65e3deb <+59>:  movl   0x10(%ebp), %eax

    0xa65e3dee <+62>:  movl   %eax, 0x24(%esi)

    0xa65e3df1 <+65>:  movl   0x14(%ebp), %eax

    0xa65e3df4 <+68>:  movl   %eax, 0x20(%esi)

    0xa65e3df7 <+71>:  movl   0x18(%ebp), %eax

    0xa65e3dfa <+74>:  movl   %eax, 0x1c(%esi)

    0xa65e3dfd <+77>:  movl   0x2c(%esi), %eax

    0xa65e3e00 <+80>:  movl   %eax, 0x18(%esi)

    0xa65e3e03 <+83>:  movl   0x28(%esi), %eax

    0xa65e3e06 <+86>:  movl   %ecx, 0x14(%esi)

    0xa65e3e09 <+89>:  movl   %eax, %ecx

    0xa65e3e0b <+91>:  movl   0x24(%esi), %eax

    0xa65e3e0e <+94>:  movl   %edx, 0x10(%esi)

    0xa65e3e11 <+97>:  movl   %eax, %edx

    0xa65e3e13 <+99>:  movl   0x20(%esi), %eax

    0xa65e3e16 <+102>: movl   %eax, %esi

->  0xa65e3e18 <+104>: movl   0x1c(%esi), %eax

    0xa65e3e1b <+107>: movl   %edi, 0xc(%esi)

    0xa65e3e1e <+110>: movl   %eax, %edi

    0xa65e3e20 <+112>: movl   0x18(%esi), %eax

    0xa65e3e23 <+115>: pushl  %ebx

    0xa65e3e24 <+116>: movl   %eax, %ebx

    0xa65e3e26 <+118>: movl   $0xac, %eax

    0xa65e3e2b <+123>: int    $0x80

    0xa65e3e2d <+125>: popl   %ebx

    0xa65e3e2e <+126>: movl   %eax, 0x38(%esi)

    0xa65e3e31 <+129>: movl   %ebx, 0x8(%esi)

    0xa65e3e34 <+132>: cmpl   $0xfffff001, 0x38(%esi)   ; imm = 0xFFFFF001 

    0xa65e3e3b <+139>: jb     0xa65e3e5d                ; <+173> at linux_syscall_support.h:3479

    0xa65e3e41 <+145>: xorl   %eax, %eax

    0xa65e3e43 <+147>: subl   0x38(%esi), %eax

    0xa65e3e46 <+150>: movl   0x34(%esi), %ebx

    0xa65e3e49 <+153>: movl   %eax, 0x4(%esi)

    0xa65e3e4c <+156>: calll  0xa65d9980                ; symbol stub for: __errno

    0xa65e3e51 <+161>: movl   0x4(%esi), %ebx

    0xa65e3e54 <+164>: movl   %ebx, (%eax)

    0xa65e3e56 <+166>: movl   $0xffffffff, 0x38(%esi)   ; imm = 0xFFFFFFFF 

    0xa65e3e5d <+173>: movl   0x38(%esi), %eax

    0xa65e3e60 <+176>: leal   -0xc(%ebp), %esp

    0xa65e3e63 <+179>: popl   %esi

    0xa65e3e64 <+180>: popl   %edi

    0xa65e3e65 <+181>: popl   %ebx

    0xa65e3e66 <+182>: popl   %ebp

    0xa65e3e67 <+183>: retl   

(lldb) register read

General Purpose Registers:

       eax = 0x00000000

       ebx = 0x00000001

       ecx = 0x00000001

       edx = 0x00000000

       edi = 0x00000000

       esi = 0x00000000

       ebp = 0xbf93e668

       esp = 0xbf93e610

       eip = 0xa65e3e18  libcrashmanager.so`sys_prctl(int, unsigned long, unsigned long, unsigned long, unsigned long) + 104 at linux_syscall_support.h:3479

    eflags = 0x00010286

        cs = 0x00000073

        fs = 0x0000003b

        gs = 0x00000033

        ss = 0x0000007b

        ds = 0x0000007b

        es = 0x0000007b

So esi being 0 is a problem obviously, but I’m having trouble following the register gymnastics leading up to it, which I guess are trying to setup the call properly.

j. 

[Jay Soffian]

Tracing through this assembly, you can see that the movl at offset +102 is what’s clobbering %esi. If you further trace back, the source of that value is argument 4 to the sys_prctl call. Indeed, if I change the call to: 

  sys_prctl(PR_SET_DUMPABLE, 1, 0, 0xee, 0);

Then %esi ends up with 0xee loaded in it. AFAIU, all the assembly before offset +115 is being generated by clang. The inline assembly in linux_syscall_support.h for _syscall5 begins with the “pushl %ebx” at offset +115.

So the question is, why is clang generating assembly which is clobbering %esi with an argument?

I’m at a bit of a loss here… this is definitely outside my $dayjob wheelhouse.

Also, how come Chromium hasn’t been bitten by this?

j.

[Mike Frysinger]

thanks for doing all of this.

what version of clang is in that toolchain ?

i've written a bunch of LSS unittests and they're all passing w/clang-5.0.  i had mmap tests, but no prctl, but i've added those now too and they pass.  however, i'm not building/running under Android.  you can check them here:

https://chromium-review.googlesource.com/446421

https://chromium-review.googlesource.com/855076

here's a simple test file that should make it easier to check the generated asm.  normally the extra args to prctl should be set to 0, but i think in the GET_DUMPABLE case they get safely ignored, and it'll help with debugging the generated asm.

$ cat test.c

#include "linux_syscall_support.h"

int main() {

  return sys_prctl(PR_GET_DUMPABLE, 11, 22, 33, 44);

}

$ clang test.c

$ ./a.out; echo $?

1

$ clang -S -o - test.c -O3

        .text

        .file   "test.c"

        .globl  main                    # -- Begin function main

        .p2align        4, 0x90

        .type   main,@function

main:                                   # @main

        .cfi_startproc

# BB#0:

        pushq   %rbx

.Lcfi0:

        .cfi_def_cfa_offset 16

.Lcfi1:

        .cfi_offset %rbx, -16

        movl    $33, %r9d    # arg4

        movl    $44, %ebx    # arg5

        movl    $3, %edi    # PR_GET_DUMPABLE

        movl    $11, %esi    # arg2

        movl    $22, %edx   # arg3

        movl    $157, %eax    # SYS_prctl

        #APP

        movq    %r9, %r10

        movq    %rbx, %r8

        syscall

        #NO_APP

        movq    %rax, %rbx

        cmpq    $-4095, %rbx            # imm = 0xF001

        jb      .LBB0_2

# BB#1:

        negl    %ebx

        callq   __errno_location

        movl    %ebx, (%rax)

        movq    $-1, %rbx

.LBB0_2:

        movl    %ebx, %eax

        popq    %rbx

        retq

.Lfunc_end0:

        .size   main, .Lfunc_end0-main

        .cfi_endproc

                                        # -- End function

-mike

--
You received this message because you are subscribed to the Google Groups "google-breakpad-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-d...@googlegroups.com.
To post to this group, send email to google-brea...@googlegroups.com.

[Jay Soffian]

If I compile with:

/Applications/android-ndk-r16b/toolchains/llvm/prebuilt/darwin-x86_64/bin/clang -gcc-toolchain /Applications/android-ndk-r16b/toolchains/x86-4.9/prebuilt/darwin-x86_64 -target i686-none-linux-android --sysroot /Applications/android-ndk-r16b/sysroot -isystem /Applications/android-ndk-r16b/sysroot/usr/include/i686-linux-android -S -o - test_lss.c -O3

.text

.file "test_lss.c"

.globl main

.p2align 4, 0x90

.type main,@function

main:                                   # @main

# BB#0:

pushl %ebp

movl %esp, %ebp

pushl %ebx

pushl %edi

pushl %esi

subl $12, %esp

calll .L0$pb

.L0$pb:

popl %ebx

.Ltmp0:

addl $_GLOBAL_OFFSET_TABLE_+(.Ltmp0-.L0$pb), %ebx

movl $11, %ecx

movl $22, %edx

movl $33, %esi

movl $44, %edi

#APP

pushl %ebx

movl $3, %ebx

movl $172, %eax

int $128

popl %ebx

#NO_APP

movl %eax, %esi

cmpl $-4095, %esi            # imm = 0xF001

jb .LBB0_2

# BB#1:

negl %esi

calll __errno@PLT

movl %esi, (%eax)

movl $-1, %esi

.LBB0_2:

movl %esi, %eax

addl $12, %esp

popl %esi

popl %edi

popl %ebx

popl %ebp

retl

.Lfunc_end0:

.size main, .Lfunc_end0-main

.ident "Android clang version 5.0.300080  (based on LLVM 5.0.300080)"

.section ".note.GNU-stack","",@progbits

That looks sane to me. If I switch to -O0:

sys_prctl:                              # @sys_prctl

# BB#0:

pushl %ebp

movl %esp, %ebp

pushl %ebx

pushl %edi

pushl %esi

subl $60, %esp

calll .L1$pb

.L1$pb:

popl %eax

.Ltmp1:

addl $_GLOBAL_OFFSET_TABLE_+(.Ltmp1-.L1$pb), %eax

movl 24(%ebp), %ecx

movl 20(%ebp), %edx

movl 16(%ebp), %esi

movl 12(%ebp), %edi

movl 8(%ebp), %ebx

movl %eax, -20(%ebp)         # 4-byte Spill

movl 8(%ebp), %eax

movl %eax, -24(%ebp)         # 4-byte Spill

movl 12(%ebp), %eax

movl %eax, -28(%ebp)         # 4-byte Spill

movl 16(%ebp), %eax

movl %eax, -32(%ebp)         # 4-byte Spill

movl 20(%ebp), %eax

movl %eax, -36(%ebp)         # 4-byte Spill

movl 24(%ebp), %eax

movl %eax, -40(%ebp)         # 4-byte Spill

movl -24(%ebp), %eax         # 4-byte Reload

movl %eax, -44(%ebp)         # 4-byte Spill

movl -28(%ebp), %eax         # 4-byte Reload

movl %ecx, -48(%ebp)         # 4-byte Spill

movl %eax, %ecx

movl -32(%ebp), %eax         # 4-byte Reload

movl %edx, -52(%ebp)         # 4-byte Spill

movl %eax, %edx

movl -36(%ebp), %eax         # 4-byte Reload

movl %esi, -56(%ebp)         # 4-byte Spill

movl %eax, %esi

movl -40(%ebp), %eax         # 4-byte Reload

movl %edi, -60(%ebp)         # 4-byte Spill

movl %eax, %edi

movl -44(%ebp), %eax         # 4-byte Reload

#APP

pushl %ebx

movl %eax, %ebx

movl $172, %eax

int $128

popl %ebx

#NO_APP

movl %eax, -16(%ebp)

movl %ebx, -64(%ebp)         # 4-byte Spill

But check out what happens when I add -mstackrealign:

sys_prctl:                              # @sys_prctl

# BB#0:

pushl %ebp

movl %esp, %ebp

pushl %ebx

pushl %edi

pushl %esi

andl $-16, %esp

subl $64, %esp

movl %esp, %esi

calll .L1$pb

.L1$pb:

popl %eax

.Ltmp1:

addl $_GLOBAL_OFFSET_TABLE_+(.Ltmp1-.L1$pb), %eax

movl 24(%ebp), %ecx

movl 20(%ebp), %edx

movl 16(%ebp), %edi

movl 12(%ebp), %ebx

movl %eax, 52(%esi)          # 4-byte Spill

movl 8(%ebp), %eax

movl %eax, 48(%esi)          # 4-byte Spill

movl 8(%ebp), %eax

movl %eax, 44(%esi)          # 4-byte Spill

movl 12(%ebp), %eax

movl %eax, 40(%esi)          # 4-byte Spill

movl 16(%ebp), %eax

movl %eax, 36(%esi)          # 4-byte Spill

movl 20(%ebp), %eax

movl %eax, 32(%esi)          # 4-byte Spill

movl 24(%ebp), %eax

movl %eax, 28(%esi)          # 4-byte Spill

movl 44(%esi), %eax          # 4-byte Reload

movl %eax, 24(%esi)          # 4-byte Spill

movl 40(%esi), %eax          # 4-byte Reload

movl %ecx, 20(%esi)          # 4-byte Spill

movl %eax, %ecx

movl 36(%esi), %eax          # 4-byte Reload

movl %edx, 16(%esi)          # 4-byte Spill

movl %eax, %edx

movl 32(%esi), %eax          # 4-byte Reload

movl %eax, %esi

movl 28(%esi), %eax          # 4-byte Reload

movl %edi, 12(%esi)          # 4-byte Spill

movl %eax, %edi

movl 24(%esi), %eax          # 4-byte Reload

#APP

pushl %ebx

movl %eax, %ebx

movl $172, %eax

int $128

popl %ebx

#NO_APP

movl %eax, 56(%esi)

movl %ebx, 8(%esi)           # 4-byte Spill

That looks to me like it clobbers %esi.

In any case, it's _not_ reproducing with this simple test case at -O1 and above. So I switched my Android build back to release mode and indeed, it gets past the sys_prctl call. But then it crashes immediately after on the memset(). Here's the dissassembly:

(lldb) dis

libcrashmanager.so`google_breakpad::ExceptionHandler::HandleSignal:

    0xa6648780 <+0>:   pushl  %ebp

    0xa6648781 <+1>:   movl   %esp, %ebp

    0xa6648783 <+3>:   pushl  %ebx

    0xa6648784 <+4>:   pushl  %edi

    0xa6648785 <+5>:   pushl  %esi

    0xa6648786 <+6>:   andl   $-0x10, %esp

    0xa6648789 <+9>:   subl   $0x10, %esp

    0xa664878c <+12>:  movl   %esp, %esi

    0xa664878e <+14>:  calll  0xa6648793                ; <+19> at exception_handler.cc:445

    0xa6648793 <+19>:  popl   %ebx

    0xa6648794 <+20>:  addl   $0x344c1, %ebx            ; imm = 0x344C1 

    0xa664879a <+26>:  movl   0x8(%ebp), %ecx

    0xa664879d <+29>:  movl   (%ecx), %eax

    0xa664879f <+31>:  testl  %eax, %eax

    0xa66487a1 <+33>:  je     0xa66487b2                ; <+50> at exception_handler.cc

    0xa66487a3 <+35>:  subl   $0xc, %esp

    0xa66487a6 <+38>:  pushl  0x8(%ecx)

    0xa66487a9 <+41>:  calll  *%eax

    0xa66487ab <+43>:  addl   $0x10, %esp

    0xa66487ae <+46>:  testb  %al, %al

    0xa66487b0 <+48>:  je     0xa66487ea                ; <+106> at linux_syscall_support.h

    0xa66487b2 <+50>:  movl   0x10(%ebp), %eax

    0xa66487b5 <+53>:  movl   0x8(%eax), %eax

    0xa66487b8 <+56>:  testl  %eax, %eax

    0xa66487ba <+58>:  jle    0xa66487f1                ; <+113> at exception_handler.cc:453

    0xa66487bc <+60>:  movl   $0x1, %ecx

    0xa66487c1 <+65>:  xorl   %edx, %edx

    0xa66487c3 <+67>:  xorl   %esi, %esi

    0xa66487c5 <+69>:  xorl   %edi, %edi

    0xa66487c7 <+71>:  pushl  %ebx

    0xa66487c8 <+72>:  movl   $0x4, %ebx

    0xa66487cd <+77>:  movl   $0xac, %eax

    0xa66487d2 <+82>:  int    $0x80

    0xa66487d4 <+84>:  popl   %ebx

    0xa66487d5 <+85>:  movl   %eax, %edi

    0xa66487d7 <+87>:  cmpl   $0xfffff001, %edi         ; imm = 0xFFFFF001 

    0xa66487dd <+93>:  jb     0xa6648807                ; <+135> at exception_handler.cc:459

    0xa66487df <+95>:  negl   %edi

    0xa66487e1 <+97>:  calll  0xa66452b0                ; symbol stub for: __errno

    0xa66487e6 <+102>: movl   %edi, (%eax)

    0xa66487e8 <+104>: jmp    0xa6648807                ; <+135> at exception_handler.cc:459

    0xa66487ea <+106>: xorl   %eax, %eax

    0xa66487ec <+108>: jmp    0xa66488b0                ; <+304> at exception_handler.cc:487

    0xa66487f1 <+113>: je     0xa66487f8                ; <+120> at exception_handler.cc:453

    0xa66487f3 <+115>: cmpl   $-0x6, %eax

    0xa66487f6 <+118>: jne    0xa6648807                ; <+135> at exception_handler.cc:459

    0xa66487f8 <+120>: movl   0x10(%ebp), %eax

    0xa66487fb <+123>: movl   0xc(%eax), %edi

    0xa66487fe <+126>: calll  0xa6645460                ; symbol stub for: getpid

    0xa6648803 <+131>: cmpl   %eax, %edi

    0xa6648805 <+133>: je     0xa66487bc                ; <+60> [inlined] sys_prctl(int, unsigned long, unsigned long, unsigned long, unsigned long) at exception_handler.cc:454

    0xa6648807 <+135>: subl   $0x4, %esp

    0xa664880a <+138>: leal   0x4b4(%ebx), %eax

    0xa6648810 <+144>: pushl  $0x158                    ; imm = 0x158 

    0xa6648815 <+149>: pushl  $0x0

    0xa6648817 <+151>: pushl  %eax

    0xa6648818 <+152>: calll  0xa66453e0                ; symbol stub for: memset

    0xa664881d <+157>: addl   $0x10, %esp

    0xa6648820 <+160>: leal   0x434(%ebx), %eax

    0xa6648826 <+166>: subl   $0x4, %esp

    0xa6648829 <+169>: pushl  $0x80

    0xa664882e <+174>: pushl  0x10(%ebp)

->  0xa6648831 <+177>: movl   %eax, 0x8(%esi)

    0xa6648834 <+180>: pushl  %eax

    0xa6648835 <+181>: calll  0xa6645260                ; symbol stub for: memcpy

    0xa664883a <+186>: addl   $0xc, %esp

    0xa664883d <+189>: leal   0x4b8(%ebx), %eax

    0xa6648843 <+195>: pushl  $0xe4

    0xa6648848 <+200>: movl   0x14(%ebp), %ecx

    0xa664884b <+203>: movl   %ecx, %edi

    0xa664884d <+205>: pushl  %edi

    0xa664884e <+206>: pushl  %eax

    0xa664884f <+207>: calll  0xa6645260                ; symbol stub for: memcpy

    0xa6648854 <+212>: addl   $0x10, %esp

    0xa6648857 <+215>: movl   0x60(%edi), %eax

    0xa664885a <+218>: testl  %eax, %eax

    0xa664885c <+220>: je     0xa6648873                ; <+243> at exception_handler.cc:479

    0xa664885e <+222>: subl   $0x4, %esp

    0xa6648861 <+225>: leal   0x59c(%ebx), %ecx

    0xa6648867 <+231>: pushl  $0x70

    0xa6648869 <+233>: pushl  %eax

    0xa664886a <+234>: pushl  %ecx

    0xa664886b <+235>: calll  0xa6645260                ; symbol stub for: memcpy

    0xa6648870 <+240>: addl   $0x10, %esp

    0xa6648873 <+243>: subl   $0xc, %esp

    0xa6648876 <+246>: pushl  $0xe0

    0xa664887b <+251>: calll  0xa6645470                ; symbol stub for: syscall

    0xa6648880 <+256>: addl   $0x10, %esp

    0xa6648883 <+259>: movl   %eax, 0x4b4(%ebx)

    0xa6648889 <+265>: movl   0x8(%ebp), %edi

    0xa664888c <+268>: cmpl   $0x0, 0x68(%edi)

    0xa6648890 <+272>: je     0xa66488b2                ; <+306> at exception_handler.cc:486

    0xa6648892 <+274>: movl   0x68(%edi), %eax

    0xa6648895 <+277>: subl   $0x4, %esp

    0xa6648898 <+280>: pushl  0x8(%edi)

    0xa664889b <+283>: pushl  $0x1d8                    ; imm = 0x1D8 

    0xa66488a0 <+288>: pushl  0x8(%esi)

    0xa66488a3 <+291>: calll  *%eax

    0xa66488a5 <+293>: addl   $0x10, %esp

    0xa66488a8 <+296>: movl   %eax, %ecx

    0xa66488aa <+298>: movb   $0x1, %al

    0xa66488ac <+300>: testb  %cl, %cl

    0xa66488ae <+302>: je     0xa66488b2                ; <+306> at exception_handler.cc:486

    0xa66488b0 <+304>: jmp    0xa66488be                ; <+318> at exception_handler.cc:487

    0xa66488b2 <+306>: subl   $0x8, %esp

    0xa66488b5 <+309>: pushl  0x8(%esi)

    0xa66488b8 <+312>: pushl  %edi

    0xa66488b9 <+313>: calll  0xa6645490                ; symbol stub for: google_breakpad::ExceptionHandler::GenerateDump(google_breakpad::ExceptionHandler::CrashContext*)

    0xa66488be <+318>: leal   -0xc(%ebp), %esp

    0xa66488c1 <+321>: popl   %esi

    0xa66488c2 <+322>: popl   %edi

    0xa66488c3 <+323>: popl   %ebx

    0xa66488c4 <+324>: popl   %ebp

    0xa66488c5 <+325>: retl   

(lldb) reg r

General Purpose Registers:

       eax = 0xa667d088  libcrashmanager.so`google_breakpad::(anonymous namespace)::g_crash_context_

       ebx = 0xa667cc54  libcrashmanager.so`_GLOBAL_OFFSET_TABLE_

       ecx = 0x00000050

       edx = 0xa667d260  libcrashmanager.so`google_breakpad::(anonymous namespace)::g_handler_stack_

       edi = 0x00000000

       esi = 0x00000000

       ebp = 0xa1f98548

       esp = 0xa1f98514

       eip = 0xa6648831  libcrashmanager.so`google_breakpad::ExceptionHandler::HandleSignal(int, siginfo*, void*) + 177 at exception_handler.cc:459

    eflags = 0x00010292

        cs = 0x00000073

        fs = 0x0000003b

        gs = 0x00000033

        ss = 0x0000007b

        ds = 0x0000007b

        es = 0x0000007b

Looking at the assembly, wait what's this:

    0xa66487c3 <+67>:  xorl   %esi, %esi   ; zero's esi, right?

...  ; nothing related to %esi occurs

->  0xa6648831 <+177>: movl   %eax, 0x8(%esi)  ; BOOM

WTF clang?

I don't know where to go with this.

j.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-discuss+unsub...@googlegroups.com.
To post to this group, send email to google-breakpad-discuss@googlegroups.com.

Visit this group at https://groups.google.com/group/google-breakpad-discuss.
For more options, visit https://groups.google.com/d/optout.

--

You received this message because you are subscribed to the Google Groups "google-breakpad-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-discuss+unsub...@googlegroups.com.
To post to this group, send email to google-breakpad-discuss@googlegroups.com.

[Mike Frysinger]

sorry, i was looking at x86_64, and you were clearly looking at x86

can you run the bad clang invocation and add -v ?  that should hopefully show what extra args are being passed in.  iirc, Android enables PIE by default now, but it sounds like it's also adding other -m flags.  so my default clang usage wouldn't line up.

if we're getting bad codegen, and it's fixed in newer clang, we could report it to Android, but it's unlikely they'd backport anything.  so we'd either have to figure out a workaround for LSS, or just mark that version of the Android NDK as bad and #error out at compile time and tell people to use a diff version :/.

-mike

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-d...@googlegroups.com.
To post to this group, send email to google-brea...@googlegroups.com.

Visit this group at https://groups.google.com/group/google-breakpad-discuss.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "google-breakpad-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-d...@googlegroups.com.
To post to this group, send email to google-brea...@googlegroups.com.

Visit this group at https://groups.google.com/group/google-breakpad-discuss.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "google-breakpad-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-d...@googlegroups.com.
To post to this group, send email to google-brea...@googlegroups.com.

[Jay Soffian]

I've updated the invocation to match how the NDK is calling clang++. Here it is at O2, which seems sane:

+ NDK=/Applications/android-ndk-r16b

+ /Applications/android-ndk-r16b/toolchains/llvm/prebuilt/darwin-x86_64/bin/clang++ -gcc-toolchain /Applications/android-ndk-r16b/toolchains/x86-4.9/prebuilt/darwin-x86_64 -target i686-none-linux-android -ffunction-sections -funwind-tables -fstack-protector-strong -fPIC -no-canonical-prefixes -fno-exceptions -fno-rtti -DNDEBUG -I/Applications/android-ndk-r16b/sources/cxx-stl/stlport/stlport -I/Applications/android-ndk-r16b/sources/cxx-stl//gabi++/include -DANDROID -D__ANDROID_API__=19 -mstackrealign -std=c++11 -frtti --sysroot /Applications/android-ndk-r16b/sysroot -isystem /Applications/android-ndk-r16b/sysroot/usr/include/i686-linux-android -S -o - test_lss.cc -O2 -v

Android clang version 5.0.300080  (based on LLVM 5.0.300080)

Target: i686-none-linux-android

Thread model: posix

InstalledDir: /Applications/android-ndk-r16b/toolchains/llvm/prebuilt/darwin-x86_64/bin

Found candidate GCC installation: /Applications/android-ndk-r16b/toolchains/x86-4.9/prebuilt/darwin-x86_64/lib/gcc/i686-linux-android/4.9.x

Selected GCC installation: /Applications/android-ndk-r16b/toolchains/x86-4.9/prebuilt/darwin-x86_64/lib/gcc/i686-linux-android/4.9.x

Candidate multilib: .;@m32

Selected multilib: .;@m32

 "/Applications/android-ndk-r16b/toolchains/llvm/prebuilt/darwin-x86_64/bin/clang++" -cc1 -triple i686-none-linux-android -S -disable-free -disable-llvm-verifier -discard-value-names -main-file-name test_lss.cc -mrelocation-model pic -pic-level 2 -mthread-model posix -fmath-errno -masm-verbose -mconstructor-aliases -munwind-tables -fuse-init-array -target-cpu i686 -target-feature +ssse3 -target-linker-version 2.24 -momit-leaf-frame-pointer -v -dwarf-column-info -debugger-tuning=gdb -ffunction-sections -coverage-notes-file /Users/jsoffian/Work/code/ycm/YCrashManager_Android/ycrashmanager/-.gcno -resource-dir /Applications/android-ndk-r16b/toolchains/llvm/prebuilt/darwin-x86_64/lib64/clang/5.0.300080 -isystem /Applications/android-ndk-r16b/sysroot/usr/include/i686-linux-android -D NDEBUG -I /Applications/android-ndk-r16b/sources/cxx-stl/stlport/stlport -I /Applications/android-ndk-r16b/sources/cxx-stl//gabi++/include -D ANDROID -D __ANDROID_API__=19 -isysroot /Applications/android-ndk-r16b/sysroot -internal-isystem /Applications/android-ndk-r16b/sysroot/usr/local/include -internal-isystem /Applications/android-ndk-r16b/toolchains/llvm/prebuilt/darwin-x86_64/lib64/clang/5.0.300080/include -internal-externc-isystem /Applications/android-ndk-r16b/sysroot/include -internal-externc-isystem /Applications/android-ndk-r16b/sysroot/usr/include -O2 -std=c++11 -fdeprecated-macro -fdebug-compilation-dir /Users/jsoffian/Work/code/ycm/YCrashManager_Android/ycrashmanager -ferror-limit 19 -fmessage-length 0 -femulated-tls -stack-protector 2 -mstackrealign -fobjc-runtime=gcc -fdiagnostics-show-option -vectorize-loops -vectorize-slp -o - -x c++ test_lss.cc

clang -cc1 version 5.0.300080 based upon LLVM 5.0.300080 default target x86_64-apple-darwin17.5.0

ignoring nonexistent directory "/Applications/android-ndk-r16b/sysroot/usr/local/include"

ignoring nonexistent directory "/Applications/android-ndk-r16b/sysroot/include"

#include "..." search starts here:

#include <...> search starts here:

 /Applications/android-ndk-r16b/sources/cxx-stl/stlport/stlport

 /Applications/android-ndk-r16b/sources/cxx-stl//gabi++/include

 /Applications/android-ndk-r16b/sysroot/usr/include/i686-linux-android

 /Applications/android-ndk-r16b/toolchains/llvm/prebuilt/darwin-x86_64/lib64/clang/5.0.300080/include

 /Applications/android-ndk-r16b/sysroot/usr/include

End of search list.

.text

.file "test_lss.cc"

.section .text.main,"ax",@progbits

.globl main

.p2align 4, 0x90

.type main,@function

main:                                   # @main

.cfi_startproc

# BB#0:

pushl %ebp

.Lcfi0:

.cfi_def_cfa_offset 8

.Lcfi1:

.cfi_offset %ebp, -8

movl %esp, %ebp

.Lcfi2:

.cfi_def_cfa_register %ebp

pushl %ebx

pushl %edi

pushl %esi

andl $-16, %esp

subl $16, %esp

movl %esp, %esi

.Lcfi3:

.cfi_offset %esi, -20

.Lcfi4:

.cfi_offset %edi, -16

.Lcfi5:

.cfi_offset %ebx, -12

calll .L0$pb

.L0$pb:

popl %ebx

.Ltmp0:

addl $_GLOBAL_OFFSET_TABLE_+(.Ltmp0-.L0$pb), %ebx

movl $33, %esi

movl $11, %ecx

movl $22, %edx

movl $44, %edi

#APP

pushl %ebx

movl $4, %ebx

movl $172, %eax

int $128

popl %ebx

#NO_APP

movl %eax, %edi

cmpl $-4095, %edi            # imm = 0xF001

jb .LBB0_2

# BB#1:

negl %edi

calll __errno@PLT

movl %edi, (%eax)

movl $-1, %edi

.LBB0_2:

movl %edi, %eax

leal -12(%ebp), %esp

popl %esi

popl %edi

popl %ebx

popl %ebp

retl

.Lfunc_end0:

.size main, .Lfunc_end0-main

.cfi_endproc

.ident "Android clang version 5.0.300080  (based on LLVM 5.0.300080)"

.section ".note.GNU-stack","",@progbits

When running it at -O0, the code gen re: %esi looks questionable to me, but I may be misreading the assembly:

+ NDK=/Applications/android-ndk-r16b

+ /Applications/android-ndk-r16b/toolchains/llvm/prebuilt/darwin-x86_64/bin/clang++ -gcc-toolchain /Applications/android-ndk-r16b/toolchains/x86-4.9/prebuilt/darwin-x86_64 -target i686-none-linux-android -ffunction-sections -funwind-tables -fstack-protector-strong -fPIC -no-canonical-prefixes -fno-exceptions -fno-rtti -DNDEBUG -I/Applications/android-ndk-r16b/sources/cxx-stl/stlport/stlport -I/Applications/android-ndk-r16b/sources/cxx-stl//gabi++/include -DANDROID -D__ANDROID_API__=19 -mstackrealign -std=c++11 -frtti --sysroot /Applications/android-ndk-r16b/sysroot -isystem /Applications/android-ndk-r16b/sysroot/usr/include/i686-linux-android -S -o - test_lss.cc -O0 -v

Android clang version 5.0.300080  (based on LLVM 5.0.300080)

Target: i686-none-linux-android

Thread model: posix

InstalledDir: /Applications/android-ndk-r16b/toolchains/llvm/prebuilt/darwin-x86_64/bin

Found candidate GCC installation: /Applications/android-ndk-r16b/toolchains/x86-4.9/prebuilt/darwin-x86_64/lib/gcc/i686-linux-android/4.9.x

Selected GCC installation: /Applications/android-ndk-r16b/toolchains/x86-4.9/prebuilt/darwin-x86_64/lib/gcc/i686-linux-android/4.9.x

Candidate multilib: .;@m32

Selected multilib: .;@m32

 "/Applications/android-ndk-r16b/toolchains/llvm/prebuilt/darwin-x86_64/bin/clang++" -cc1 -triple i686-none-linux-android -S -disable-free -disable-llvm-verifier -discard-value-names -main-file-name test_lss.cc -mrelocation-model pic -pic-level 2 -mthread-model posix -mdisable-fp-elim -fmath-errno -masm-verbose -mconstructor-aliases -munwind-tables -fuse-init-array -target-cpu i686 -target-feature +ssse3 -target-linker-version 2.24 -v -dwarf-column-info -debugger-tuning=gdb -ffunction-sections -coverage-notes-file /Users/jsoffian/Work/code/ycm/YCrashManager_Android/ycrashmanager/-.gcno -resource-dir /Applications/android-ndk-r16b/toolchains/llvm/prebuilt/darwin-x86_64/lib64/clang/5.0.300080 -isystem /Applications/android-ndk-r16b/sysroot/usr/include/i686-linux-android -D NDEBUG -I /Applications/android-ndk-r16b/sources/cxx-stl/stlport/stlport -I /Applications/android-ndk-r16b/sources/cxx-stl//gabi++/include -D ANDROID -D __ANDROID_API__=19 -isysroot /Applications/android-ndk-r16b/sysroot -internal-isystem /Applications/android-ndk-r16b/sysroot/usr/local/include -internal-isystem /Applications/android-ndk-r16b/toolchains/llvm/prebuilt/darwin-x86_64/lib64/clang/5.0.300080/include -internal-externc-isystem /Applications/android-ndk-r16b/sysroot/include -internal-externc-isystem /Applications/android-ndk-r16b/sysroot/usr/include -O0 -std=c++11 -fdeprecated-macro -fdebug-compilation-dir /Users/jsoffian/Work/code/ycm/YCrashManager_Android/ycrashmanager -ferror-limit 19 -fmessage-length 0 -femulated-tls -stack-protector 2 -mstackrealign -fobjc-runtime=gcc -fdiagnostics-show-option -o - -x c++ test_lss.cc

clang -cc1 version 5.0.300080 based upon LLVM 5.0.300080 default target x86_64-apple-darwin17.5.0

ignoring nonexistent directory "/Applications/android-ndk-r16b/sysroot/usr/local/include"

ignoring nonexistent directory "/Applications/android-ndk-r16b/sysroot/include"

#include "..." search starts here:

#include <...> search starts here:

 /Applications/android-ndk-r16b/sources/cxx-stl/stlport/stlport

 /Applications/android-ndk-r16b/sources/cxx-stl//gabi++/include

 /Applications/android-ndk-r16b/sysroot/usr/include/i686-linux-android

 /Applications/android-ndk-r16b/toolchains/llvm/prebuilt/darwin-x86_64/lib64/clang/5.0.300080/include

 /Applications/android-ndk-r16b/sysroot/usr/include

End of search list.

.text

.file "test_lss.cc"

.section .text.main,"ax",@progbits

.globl main

.p2align 4, 0x90

.type main,@function

main:                                   # @main

.cfi_startproc

# BB#0:

pushl %ebp

.Lcfi0:

.cfi_def_cfa_offset 8

.Lcfi1:

.cfi_offset %ebp, -8

movl %esp, %ebp

.Lcfi2:

.cfi_def_cfa_register %ebp

pushl %ebx

pushl %edi

pushl %esi

andl $-16, %esp

subl $48, %esp

.Lcfi3:

.cfi_offset %esi, -20

.Lcfi4:

.cfi_offset %edi, -16

.Lcfi5:

.cfi_offset %ebx, -12

calll .L0$pb

.L0$pb:

popl %eax

.Ltmp0:

addl $_GLOBAL_OFFSET_TABLE_+(.Ltmp0-.L0$pb), %eax

movl $4, %ecx

movl $11, %edx

movl $22, %esi

movl $33, %edi

movl $44, %ebx

movl $0, 40(%esp)

movl $4, (%esp)

movl $11, 4(%esp)

movl $22, 8(%esp)

movl $33, 12(%esp)

movl $44, 16(%esp)

movl %ebx, 36(%esp)          # 4-byte Spill

movl %eax, %ebx

movl %edi, 32(%esp)          # 4-byte Spill

movl %ecx, 28(%esp)          # 4-byte Spill

movl %edx, 24(%esp)          # 4-byte Spill

movl %esi, 20(%esp)          # 4-byte Spill

calll _ZL9sys_prctlimmmm

leal -12(%ebp), %esp

popl %esi

popl %edi

popl %ebx

popl %ebp

retl

.Lfunc_end0:

.size main, .Lfunc_end0-main

.cfi_endproc

.section .text._ZL9sys_prctlimmmm,"ax",@progbits

.p2align 4, 0x90

.type _ZL9sys_prctlimmmm,@function

_ZL9sys_prctlimmmm:                     # @_ZL9sys_prctlimmmm

.cfi_startproc

# BB#0:

pushl %ebp

.Lcfi6:

.cfi_def_cfa_offset 8

.Lcfi7:

.cfi_offset %ebp, -8

movl %esp, %ebp

.Lcfi8:

.cfi_def_cfa_register %ebp

pushl %ebx

pushl %edi

pushl %esi

andl $-16, %esp

subl $64, %esp

movl %esp, %esi

.Lcfi9:

.cfi_offset %esi, -20

.Lcfi10:

.cfi_offset %edi, -16

.Lcfi11:

.cfi_offset %ebx, -12

# BB#1:

cmpl $-4095, 56(%esi)        # imm = 0xF001

jb .LBB1_3

# BB#2:

xorl %eax, %eax

subl 56(%esi), %eax

movl 52(%esi), %ebx          # 4-byte Reload

movl %eax, 4(%esi)           # 4-byte Spill

calll __errno@PLT

movl 4(%esi), %ebx           # 4-byte Reload

movl %ebx, (%eax)

movl $-1, 56(%esi)

.LBB1_3:

movl 56(%esi), %eax

leal -12(%ebp), %esp

popl %esi

popl %edi

popl %ebx

popl %ebp

retl

.Lfunc_end1:

.size _ZL9sys_prctlimmmm, .Lfunc_end1-_ZL9sys_prctlimmmm

.cfi_endproc

.ident "Android clang version 5.0.300080  (based on LLVM 5.0.300080)"

.section ".note.GNU-stack","",@progbits

The part that looks suspicious to me is this:

movl 20(%ebp), %eax

movl %eax, 32(%esi)          # 4-byte Spill

       ...

movl 32(%esi), %eax          # 4-byte Reload

movl %eax, %esi

Why load from memory just to immediately spill it? But never mind that, then it clobbers esi with esi+0x32, which is whatever was at ebp+0x20. But looking up above, clang clearly thinks that ebp+0x20 is the fourth sys_prctl argument.

So that's the crash when compiling with APP_OPTIM=debug (which is -O0). 

The crash I'm seeing with APP_OPTIM=release (which is -O2) happens after sys_prctl returns. But if -O0 is broken I don't have confidence in -O2. Both crashes are related to esi, so it seems like make clang isn't tracking that register correctly.

Any ideas?

For now, I've switched back to building my app with GCC, so this isn't urgent. Also, I tested with the latest beta NDK and the same issue occurs. Can we loop in someone from the NDK team? Should I file an Android bug?

j.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-discuss+unsub...@googlegroups.com.
To post to this group, send email to google-breakpad-discuss@googlegroups.com.

Visit this group at https://groups.google.com/group/google-breakpad-discuss.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "google-breakpad-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-discuss+unsub...@googlegroups.com.
To post to this group, send email to google-breakpad-discuss@googlegroups.com.

Visit this group at https://groups.google.com/group/google-breakpad-discuss.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "google-breakpad-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-discuss+unsub...@googlegroups.com.
To post to this group, send email to google-breakpad-discuss@googlegroups.com.

Visit this group at https://groups.google.com/group/google-breakpad-discuss.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "google-breakpad-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to google-breakpad-discuss+unsub...@googlegroups.com.
To post to this group, send email to google-breakpad-discuss@googlegroups.com.

[Mike Frysinger]

filing a bug for the Android team is prob your best bet.  i don't have any contacts with the respective teams.

-mike

[heirish lee]

Reply here in case you still have this problem.

I met the same issue these days.

built my sdk with breakpad using android ndk r16b.

and it crashes on android x86.

fixed by using the latest lss head file, 

seems they have fixed this issue in https://chromium.googlesource.com/linux-syscall-support/+/be2d5a80df9a9519cab306508ce902d774da76a9

FYI