Sigh. No, I missed one thing. If application use
makecontext()/swapcontext() pair,
ESP is not reliable way to detect pthread stack. At that time the
stack is still marked
as anonymous memory.
This is not wrong, because it essentially gives the correct picture of
the state of that task -- the task is using another vma as a stack
during that point and not the one it was allotted by pthreads during
thread creation.
I don't think we can successfully stick to the idea of trying to mark
stack space allocated by pthreads but not used by any task *currently*
as stack as long as the allocation happens outside the kernel space.
The only way to mark this is either by marking the stack as
VM_GROWSDOWN (which will make the stack grow and break some pthreads
functions) or create a new flag, which a simple display such as this
does not deserve. So it's best that this sticks to what the kernel
*knows* is being used as stack.
Oh, maybe generically you are right. but you missed one thing. Before
your patch, stack or not stack are address space property. thus, using
/proc/pid/maps makes sense. but after your patch, it's no longer memory
property. applications can use heap or mapped file as a stack. then, at
least, current your code is wrong. the code assume each memory property
are exclusive.
Moreover, if pthread stack is unimportant, I wonder why we need this patch
at all. Which application does need it? and When?
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>