On 09/12/2017 09:47 PM, Konstantin Tokarev wrote:
13.09.2017, 04:40, "Ed Leaver" <[email protected]>:
What??? You mean there's actually a reason people aren't knocking the doors
down over these things? =-O
A few months ago I was handed a C++ Coding Standard that deigned to prohibit
any further heap allocation after program initialization. It was originally
intended for a hard real-time system; I racked my feeble mind trying to
ascertain how it could possibly apply to our particular project, and finally
suggested -- much as yourself -- that if memory allocation were felt to be a
legitimate concern, there were hooks enough described in man 5 proc to settle
the issue one way or another.
Then tactfully suggested a more appropriate standard. Tactfully enough that the
suggestion was accepted.
I ran across TCMalloc in the process. It sounds Really Neat -- but it wasn't
clear why it's Really Neat properties hadn't already been incorporated into the
standard kernel allocator. Is why I asked.
Don't confuse kernel allocator with libc (malloc) allocator.
Thanks. I had intended "system allocator" to distinguish from those used
in the kernel, but obviously did not do so.
Thanks!
On 09/12/2017 08:43 AM, Thiago Macieira wrote:
On Monday, 11 September 2017 17:45:01 PDT Ed Leaver wrote:
Have any of you experience with jemalloc or TCMalloc?
http://goog-perftools.sourceforge.net/doc/tcmalloc.html
Yes. I don't remember which of the two allocators or the details, but I
remember one of them had a huge thread-safety problem and would corrupt itself.
Use at your own risk. I will close any bug reports filed if you change malloc.
,
_______________________________________________
Development mailing list
[email protected]
http://lists.qt-project.org/mailman/listinfo/development
--
Regards,
Konstantin
_______________________________________________
Development mailing list
[email protected]
http://lists.qt-project.org/mailman/listinfo/development
