I didn't notice the UMA_ZONE_NOFREE flag of proc_zone, so proc items will not
be recycled.
But the existence of proc_fini really confused me.
Another question?
Can memory management system utilize COW to supply zero-filled page to kernel
or user process.
That is to say:
When processes want zeroed page, we give them a mirror of one already zerod
pages. If they just
read, they can read zero from the back page.
We needn't really alloc a new zero-filled page until they modify the page.
So we can saving many time from filling pages with zero, if some process just
want read from them.
Can this suggestion work?
>On Friday 21 October 2005 04:32 pm, David Schultz wrote:
>> On Fri, Oct 21, 2005, John Baldwin wrote:
>> > On Friday 21 October 2005 09:13 am, nocool wrote:
>> > > freebsd-hackers�����hello
>> > >
>> > > Question about 5.4 kernel source code.
>> > > I have some question about strust proc's initialize. Kernel use
>> > > proc_zone to allocate proc items and initialize them with proc_init
>> > > (sys\kern\kern_proc.c) function. In this function, we can find the
>> > > field proc.p_stats is allocated with pstats_alloc(), as
>> > >
>> > > p->p_stats = pstats_alloc();
>> > >
>> > > and pstats_alloc is realized as
>> > >
>> > > malloc(sizeof(struct pstats), M_SUBPROC, M_ZERO|M_WAITOK);
>> > >
>> > > But I can't find where this field is freed. If it will not be release,
>> > > will there be memory leakage?
>> >
>> > Heh, das@ forgot to call pstats_free() when he did the changes. The
>> > reason is probably because proc_fini() doesn't do anything useful because
>> > we never recycle proc structs. We should probably at least add the
>> > operations there though for documentation purposes. Something like this
>> > would work I think:
>>
>> I didn't put in the call because we never free proc structures, but
>> documenting what should happen if we ever do free them is a good
>> idea. There's a fair amount of other cleanup that needs to happen
>> as well, which you can probably find in the CVS history. (IIRC,
>> I'm guilty of removing the code at a time when more things depended
>> upon struct proc being type safe. Are there any remaining reasons
>> why we can't free struct procs at this point?)
>>
>> By the way, there's no reason why we can't fold struct pstats into
>> struct proc so we don't have to allocate and free it at all.
>> It's never shared, so the extra level of indirection just adds overhead.
>> The main reason I didn't make this change earlier was to maintain binary
>> compatibility when I backported my U-area changes to -STABLE.
>
>Looks like some of the functions (vm_dispose_proc() and sched_destroyproc())
>have vanished, so this is all that would be in there now:
>
>Index: kern_proc.c
>===================================================================
>RCS file: /usr/cvs/src/sys/kern/kern_proc.c,v
>retrieving revision 1.232
>diff -u -r1.232 kern_proc.c
>--- kern_proc.c 2 Oct 2005 23:27:56 -0000 1.232
>+++ kern_proc.c 21 Oct 2005 21:21:45 -0000
>@@ -196,8 +196,17 @@
> static void
> proc_fini(void *mem, int size)
> {
>+#ifdef notnow
>+ struct proc *p;
>
>+ p = (struct proc *)mem;
>+ pstats_free(p->p_stats);
>+ ksegrp_free(FIRST_KSEGRP_IN_PROC(p));
>+ thread_free(FIRST_THREAD_IN_PROC(p));
>+ mtx_destroy(&p->p_mtx);
>+#else
> panic("proc reclaimed");
>+#endif
> }
>
> /*
>
>
>--
>John Baldwin <[EMAIL PROTECTED]> <>< http://www.FreeBSD.org/~jhb/
>"Power Users Use the Power to Serve" = http://www.FreeBSD.org
= = = = = = = = = = = = = = = = = = = =
致
礼！
nocool
[EMAIL PROTECTED]
2005-10-22
_______________________________________________
[email protected] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-hackers
To unsubscribe, send any mail to "[EMAIL PROTECTED]"
