Re: Bizarre x86 segment layout problem causing select() slowdown: solved, sort of
Re: Bizarre x86 segment layout problem causing select() slowdown: solved, sort of
- Subject: Re: Bizarre x86 segment layout problem causing select() slowdown: solved, sort of
- From: Dave Hayden <email@hidden>
- Date: Wed, 7 Feb 2007 11:22:38 -0800
On Feb 7, 2007, at 10:56 AM, glenn andreas wrote:
On Feb 7, 2007, at 12:49 PM, Jonas Maebe wrote:
setsockopt(sock, SOL_SOCKET, SO_RCVBUF, (void*)32768, sizeof(int));
Shouldn't you pass the address of a variable containing the buffer
size rather than the actual value? Have you tried checking the
return value of that call?
It's still a bug of course if behaviour randomly changes depending
on which erroneous parameter you pass (maybe the kernel doesn't
deal gracefully with a wrong parameter there and uses an
uninitialised value in case it's invalid).
32768 ends up being smack in the middle of the __TEXT segment, so
whatever is at that address is part of the program. Thus adding
dummy routines will change what part of the program that points to,
and thus alter the behavior.
Yeah, and when it's at that "magic" value, the contents of (int*)
32768 is 583, what I was seeing returned from read().
Well, I'm going to go crawl under a rock now.. Thanks for your help,
everyone.
-D
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Xcode-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden