z*******w
发帖数: 79 | 1 I have a program with 2 threads running on Solaris using g++.
I noticed that I got core-dumped at one thread's:
new char[2];
gdb trace tell me it's inside malloc.
And I checked my program, it happend when one thread 1 send data
to another thread 2. In thread 1 it will delete some data, thread 2
will use new to create an instance.
After I comment the delete of thread 1, I got no core-dump at all.
So, I am very suspicious of g++'s new/delete is not thread safe.
Can somebody verify this? Th | m*****e
发帖数: 4193 | 2 Haven't you just verified it?
I don't think any basic C++ operations are thread safe. It's your
business, not compiler's.
【在 z*******w 的大作中提到】
: I have a program with 2 threads running on Solaris using g++.
: I noticed that I got core-dumped at one thread's:
: new char[2];
: gdb trace tell me it's inside malloc.
: And I checked my program, it happend when one thread 1 send data
: to another thread 2. In thread 1 it will delete some data, thread 2
: will use new to create an instance.
: After I comment the delete of thread 1, I got no core-dump at all.
: So, I am very suspicious of g++'s new/delete is not thread safe.
: Can somebody verify this? Th
| l******n
发帖数: 3 | 3 Not Java, no monitor, implement lock, cv by u self
【在 m*****e 的大作中提到】
: Haven't you just verified it?
: I don't think any basic C++ operations are thread safe. It's your
: business, not compiler's.
| n********g
发帖数: 10 | 4 try add -D _REENTRANT whlie compiling.
【在 z*******w 的大作中提到】
: I have a program with 2 threads running on Solaris using g++.
: I noticed that I got core-dumped at one thread's:
: new char[2];
: gdb trace tell me it's inside malloc.
: And I checked my program, it happend when one thread 1 send data
: to another thread 2. In thread 1 it will delete some data, thread 2
: will use new to create an instance.
: After I comment the delete of thread 1, I got no core-dump at all.
: So, I am very suspicious of g++'s new/delete is not thread safe.
: Can somebody verify this? Th
|
|
