-
Notifications
You must be signed in to change notification settings - Fork 566
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
APP CRASH, HANG running jdk8 apps with traces enabled #3586
Comments
I see #2506 from 2017 with similar issue on java and node apps. |
This crash looks different from the one in #2506 in that this is a SIGILL.
|
Following is another crash seen in the java app: <(1+x) Handling our fault in a TRY at 0x00007f1e22405d7c> |
simple java -version failed. <(1+x) Handling our fault in a TRY at 0x00007f8f1f679d7c> |
From the
Which is consistent with the Did you build the The same error appears in issue #3610 |
Please look in the thread log file (you've only supplied the global log file) to see what the invalid opcode is. Is it AVX-512 (#1312) whose support is in-progress? |
#java -version shouldn't execute avx512 opcodes. This is jvm built from the source, also system is x86_64. |
Please check anyway.
> This is jvm built from the source, also system is x86_64.
Could still be AVX-512. Have a look whether the instruction begins with
0x62.
*From: *rkgithubs <notifications@github.com>
*Date: *Thu, May 16, 2019 at 10:05 AM
*To: *DynamoRIO/dynamorio
*Cc: *Subscribed
#java -version shouldn't execute avx512 opcodes. This is jvm built from the
… source, also system is x86_64.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#3586?email_source=notifications&email_token=AKK7Q5KCRYBC7Z27LUGVLI3PVWH4VA5CNFSM4HJVHVE2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODVSOLNI#issuecomment-493151669>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AKK7Q5L7V635252ATRSR7OTPVWH4VANCNFSM4HJVHVEQ>
.
|
Is there a recommended version of JDK on x64. Currently all java program crash with DR for JDK9+ |
with jdk8 debug level 3; java app is stuck <proactive reset @ fragments>>. I have tried running SPEC jvm2008 and some multi-threaded tests- applications don't run/start as expected. Please suggest if any more info is needed. This is x86_64 and jdk8 built from source. ~/rahul/java_tests# /root/rahul/DynamoRIO-x86_64-Linux-7.90.18019-0/bin64/drrun -debug -loglevel 3 -- /root/rahul/jdk1.8.0_201/bin/java TestThread <curiosity: rex.w on OPSZ_6_irex10_short4!> <proactive reset @ 18507 fragments> |
top of stack trace A fatal error has been detected by the Java Runtime Environment: Internal Error (sharedRuntime.cpp:549), pid=186643, tid=0x00007f78d61f3700 JRE version: Java(TM) SE Runtime Environment (8.0_201-b09) (build 1.8.0_201-b09) If you would like to submit a bug report, please visit: --------------- T H R E A D --------------- Current thread (0x00007f79f8173000): JavaThread "3" [_thread_in_Java, id=186719, stack(0x00007f78d60f3000,0x00007f78d61f4000)] Stack: [0x00007f78d60f3000,0x00007f78d61f4000], sp=0x00007f78d61ef020, free space=1008k |
We are still waiting for info about the invalid opcode...note that Status-NeedInfo label. |
The program crashes on jdk9 due to invalid opcode. log attached SYSLOG_WARNING: Invalid opcode encountered |
although there is different crash with jdk 8. attached thread level log for this crash. SYSLOG_ERROR: Application /root/rahul/jdk1.8.0_201/bin/java (191849). Internal Error: DynamoRIO debug check failure: /home/travis/build/DynamoRIO/dynamorio/core/dispatch.c:757 wherewasi == DR_WHERE_FCACHE || wherewasi == DR_WHERE_TRAMPOLINE || wherewasi == DR_WHERE_APP || (dcontext->go_native && wherewasi == DR_WHERE_DISPATCH) |
That is AVX-512. |
The wherewasi assert is #3175. It's not serious -- maybe re-run with |
These seem surprisingly early and frequent: that much memory shouldn't be used so quickly. I would try with |
application can be run via DR after adding -disable_traces. Wouldn't it disable the trace collection completely though. Our goal is to run drcachesim tool to collect the memory access instructions. /rahul/java_tests# /root/rahul/DynamoRIO-x86_64-Linux-7.90.18019-0/bin64/drrun -s 60 -vm_size 1G -no_enable_reset -disable_traces -- /root/rahul/jdk1.8.0_201/bin/java TestThread |
The DR core's traces have no relation to drcachesim's memory access traces. See the |
once we tried to collect traces -t drachesim -offline trace collection failed with the following:
[stdout]
A fatal error has been detected by the Java Runtime Environment:
[error occurred during error reporting , id 0xb] |
(Triple backtics delimit code blocks to make it easier to read -- I added some) |
So the status is that jdk8 on even small Java apps seems to hit an app crash when run under plain DR with traces enabled, or when run under drcachesim -offline? Likely both are related to cache consistency: probably the same or similar to #2989. |
crash is seen with drcachesim -offline on small apps even.
Rahul
…On Fri, May 24, 2019, 07:25 Derek Bruening ***@***.***> wrote:
So the status is that jdk8 on even small Java apps seems to hit an app
crash when run under plain DR with traces enabled, or when run under
drcachesim -offline? Likely both are related to cache consistency: probably
the same or similar to #2989
<#2989>.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#3586?email_source=notifications&email_token=AK4GG5AZ7QNM2AD4TH6EP3DPW73FXA5CNFSM4HJVHVE2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWFQ6PA#issuecomment-495652668>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AK4GG5DTXKRUCVG7FZEU3EDPW73FXANCNFSM4HJVHVEQ>
.
|
That is surprising given regression tests and usage elsewhere. This sounds like a separate issue from this issue here about JDK8. Please file a separate issue and provide the app where this happens. |
I'm seeing the following crashes on aarch64 while running a couple of java applications, this happens to on JDK8 and JDK9 both.
First one is SPECjvm08 benchmark, second one is a physics game test.
~/workloads/SPECjvm2008/SPECjvm2008_v101$ ~/tools/DynamoRIO/DynamoRIO-x86_64-Linux-7.90.17998-0/bin64/drrun -t drcachesim -offline -- ~/tools/jdk1.9-b160/bin/java -jar SPECjvm2008.jar scimark.sor.small -ikv -wt 30 -it 60
A fatal error has been detected by the Java Runtime Environment:
SIGILL (0x4) at pc=0x00007f9541ca5383, pid=29922, tid=29923
JRE version: (9.0+160) (build )
Java VM: Java HotSpot(TM) 64-Bit Server VM (9-ea+160, mixed mode, tiered, compressed oops, g1 gc, linux-amd64)
Problematic frame:
v ~BufferBlob::Signature Handler Temp Buffer
No core dump will be written. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
An error report file with more information is saved as:
/home/daenerys/workloads/SPECjvm2008/SPECjvm2008_v101/hs_err_pid29922.log
Loaded disassembler from /home/daenerys/tools/jdk1.9-b160/lib/server/hsdis-amd64.so
If you would like to submit a bug report, please visit:
http://bugreport.java.com/bugreport/crash.jsp
~/workloads/SPECjvm2008/SPECjvm2008_v101$ ~/tools/DynamoRIO/DynamoRIO-x86_64-Linux-7.90.17998-0/bin64/drrun -t drcachesim -offline -- ~/tools/jdk1.9-b160/bin/java -jar ~/workloads/IcyRocks-Java-Kernel-V1.0.2.jar
A fatal error has been detected by the Java Runtime Environment:
SIGILL (0x4) at pc=0x00007f1c11ca5383, pid=29991, tid=29992
JRE version: (9.0+160) (build )
Java VM: Java HotSpot(TM) 64-Bit Server VM (9-ea+160, mixed mode, tiered, compressed oops, g1 gc, linux-amd64)
Problematic frame:
v ~BufferBlob::Signature Handler Temp Buffer
No core dump will be written. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
An error report file with more information is saved as:
/home/daenerys/workloads/SPECjvm2008/SPECjvm2008_v101/hs_err_pid29991.log
Loaded disassembler from /home/daenerys/tools/jdk1.9-b160/lib/server/hsdis-amd64.so
If you would like to submit a bug report, please visit:
http://bugreport.java.com/bugreport/crash.jsp
Segmentation fault
The text was updated successfully, but these errors were encountered: