[jira] [Resolved] (IMPALA-9464) Tuple::DeepCopyVarlenData crashes on memcpy

["Tim Armstrong (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/IMPALA-9464?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Tim Armstrong resolved IMPALA-9464.
-----------------------------------
    Resolution: Cannot Reproduce

I don't think we have enough info to take action.

> Tuple::DeepCopyVarlenData crashes on memcpy
> -------------------------------------------
>
>                 Key: IMPALA-9464
>                 URL: https://issues.apache.org/jira/browse/IMPALA-9464
>             Project: IMPALA
>          Issue Type: Bug
>            Reporter: Norbert Luksa
>            Priority: Critical
>
> The stack trace is
> {code:java}
> Thread 1 (LWP 29440):
> [...]
> #6 <signal handler called>
> #7 0x00000037e4e896ce in memcpy () from ./lib64/libc.so.6
> #8 0x0000000000b3fcc6 in impala::Tuple::DeepCopyVarlenData (this=0x7f318b494948, desc=..., pool=pool@entry=0x7f3486927468) at /usr/src/debug/impala-2.11.0-cdh5.14.2/be/src/runtime/tuple.cc:103
> #9 0x0000000000b3fdf4 in impala::Tuple::DeepCopy (this=this@entry=0x7f31f0b200f6, dst=dst@entry=0x7f318b494948, desc=..., pool=pool@entry=0x7f3486927468) at /usr/src/debug/impala-2.11.0-cdh5.14.2/be/src/runtime/tuple.cc:92
> #10 0x0000000000b3fe8f in impala::Tuple::DeepCopy (this=0x7f31f0b200f6, desc=..., pool=0x7f3486927468) at /usr/src/debug/impala-2.11.0-cdh5.14.2/be/src/runtime/tuple.cc:83
> #11 0x00000000011e5811 in impala::DataStreamSender::Channel::AddRow (this=0x7f349da09500, row=row@entry=0x191a7f80) at /usr/src/debug/impala-2.11.0-cdh5.14.2/be/src/runtime/data-stream-sender.cc:258
> #12 0x00000000011e63af in impala::DataStreamSender::Send (this=0x7f3485ca7860, state=0x21b32e80, batch=0x262ae750) at /usr/src/debug/impala-2.11.0-cdh5.14.2/be/src/runtime/data-stream-sender.cc:475
> #13 0x0000000000b72124 in impala::FragmentInstanceState::ExecInternal (this=this@entry=0x1a4ad680) at /usr/src/debug/impala-2.11.0-cdh5.14.2/be/src/runtime/fragment-instance-state.cc:275
> #14 0x0000000000b749f2 in impala::FragmentInstanceState::Exec (this=this@entry=0x1a4ad680) at /usr/src/debug/impala-2.11.0-cdh5.14.2/be/src/runtime/fragment-instance-state.cc:89
> #15 0x0000000000b64038 in impala::QueryState::ExecFInstance (this=0x14347600, fis=0x1a4ad680) at /usr/src/debug/impala-2.11.0-cdh5.14.2/be/src/runtime/query-state.cc:382
> [..]
> {code}
> Relevant part of Tuple::DeepCopyVarlenData
> {code:c++}
> char* string_copy = reinterpret_cast<char*>(pool->Allocate(string_v->len));
> Ubsan::MemCpy(string_copy, string_v->ptr, string_v->len);
> {code}
> We could rewrite the function to return Status. This does not solve the root cause, but at least we could avoid crashes if the allocation fails.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)