Skip to contentSkip to conversation

PG_slab?

DISCUSSION Marcelo Tosatti 3 participants 3 messages Sep 20, 2004 – Sep 21, 2004 Memory management

lore
Share & tools
lore
#0 lore raw

Hi MM fellows,

What is PG_slab about?

#define PG_slab 7 /* slab debug (Suparna wants this) */

Its not used by SLAB though:

[marcelo@xeon mm]$ grep -A5 -B5 PG_slab page_alloc.c

                        1 << PG_lru     |
                        1 << PG_private |
                        1 << PG_locked  |
                        1 << PG_active  |
                        1 << PG_reclaim |
                        1 << PG_slab    |
                        1 << PG_swapcache |
                        1 << PG_writeback )))
                bad_page(function, page);
        if (PageDirty(page))
                ClearPageDirty(page);

[marcelo@xeon mm]$

I suppose it was used by someone sometime ago but its not anymore?

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#1 lore raw
Quoted 1 line · Marcelo

What is PG_slab about?

A grep in a 2.6.0-mm1 I have expanded shows that its some slab debug stuff that Suparna wanted. The last grep at the bottom of this display is in the routine mm/slab.c: ptrinfo(), that dumps data about some address.

arch/arm/mm/init.c- cached++;
arch/arm/mm/init.c: else if (PageSlab(page))
arch/arm/mm/init.c- slab++;
--
arch/arm26/mm/init.c- cached++;
arch/arm26/mm/init.c: else if (PageSlab(page))
arch/arm26/mm/init.c- slab++;
include/linux/page-flags.h-#define PG_active 6
include/linux/page-flags.h:#define PG_slab 7 /* slab debug (Suparna wants this) */
include/linux/page-flags.h-
--
include/linux/page-flags.h-
include/linux/page-flags.h:#define PageSlab(page) test_bit(PG_slab, &(page)->flags)
include/linux/page-flags.h:#define SetPageSlab(page) set_bit(PG_slab, &(page)->flags)
include/linux/page-flags.h:#define ClearPageSlab(page) clear_bit(PG_slab, &(page)->flags)
include/linux/page-flags.h:#define TestClearPageSlab(page) test_and_clear_bit(PG_slab, &(page)->flags)
include/linux/page-flags.h:#define TestSetPageSlab(page) test_and_set_bit(PG_slab, &(page)->flags)
include/linux/page-flags.h-
mm/nommu.c-
mm/nommu.c: if (PageSlab(page))
mm/nommu.c- return ksize(objp);
--
mm/page_alloc.c- 1 << PG_reclaim |
mm/page_alloc.c: 1 << PG_slab |
mm/page_alloc.c- 1 << PG_writeback )))
--
mm/slab.c- add_page_state(nr_slab, i);
mm/slab.c- while (i--) {
mm/slab.c: SetPageSlab(page);
mm/slab.c- page++;
mm/slab.c- }
--
mm/slab.c-
mm/slab.c- while (i--) {
mm/slab.c: if (!TestClearPageSlab(page))
mm/slab.c- BUG();
mm/slab.c- page++;
--
mm/slab.c- }
mm/slab.c- page = virt_to_page(objp);
mm/slab.c: if (!PageSlab(page)) {
mm/slab.c- printk(KERN_ERR "kfree_debugcheck: bad ptr %lxh.\n", (unsigned long)objp);
mm/slab.c- BUG();
--
mm/slab.c- page = virt_to_page((void*)addr);
mm/slab.c- printk("struct page at %p, flags %lxh.\n", page, page->flags);
mm/slab.c: if (PageSlab(page)) {
mm/slab.c- kmem_cache_t *c;
mm/slab.c- struct slab *s;

Signature · 8 lines
                          I won't rest till it's the best ...
                          Programmer, Linux Scalability
                          Paul Jackson <[email protected]> 1.650.933.1373
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#2 lore raw
Quoted 7 lines · Paul Jackson
Quoted 1 line · Marcelo

What is PG_slab about?

A grep in a 2.6.0-mm1 I have expanded shows that its some slab debug stuff that Suparna wanted. The last grep at the bottom of this display is in the routine mm/slab.c: ptrinfo(), that dumps data about some address.

I guess this patch or something like it was merged:

Date: Wed, 1 Sep 2004 12:02:23 +0200
From: Christoph Hellwig <[email protected]>
Subject: [PATCH] remove ptrinfo

http://lkml.org/lkml/2004/9/1/70
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>