On Thu, 21 Feb 2013, Shaohua Li wrote:
> In swapin_readahead(), read_swap_cache_async() can read a bad swap entry,
> because we don't check if readahead swap entry is bad. This doesn't break
> anything but such swapin page is wasteful and can only be freed at page
> reclaim. We avoid read such swap entry.
>
> And next patch will mark a swap entry bad temporarily for discard. Without this
> patch, swap entry count will be messed.
>
> Thanks Hugh to inspire swapin_readahead could use bad swap entry.
>
> Signed-off-by: Shaohua Li <shli@xxxxxxxxxxxx>
Acked-by: Hugh Dickins <hughd@xxxxxxxxxx>
Personally, I'd have merged this one into the next, or added it just
after the next - it can be easier to explain a bug once it's already
there, and okay to fix up after, just so long as it's very unlikely to
interfere with anybody's future bisection. But maybe you and I just
prefer to tell stories in a different way: no need for you to reorder.
I applied your v3 series shortly before 3.9-rc1, and ran it under load
for a week. With this addition to your previous, I no longer saw any
"Unused swap" or VM_BUG_ON(error == -EEXIST) issues, it all ran fine.
With one exception: swapoff occasionally failed, and it was another
such SWAP_MAP_BAD issue. At the bottom I've appended the patch I was
using to fix that (which of course only makes sense after your next).
Maybe you'd like to merge that into this 2/4, and exchange 2/4 and 3/4,
or maybe you'd just like to merge it into your 3/4, or maybe you'd
prefer to keep it as a separate fixup following 3/4: up to you, I'm
not hung up the the ownership of it. (I never experimented without
the SWP_WRITEOK part of it: I'm not sure how necessary that part is,
but I feel safer with it in.)
> ---
> mm/swapfile.c | 5 +++++
> 1 file changed, 5 insertions(+)
>
> Index: linux/mm/swapfile.c
> ===================================================================
> --- linux.orig/mm/swapfile.c 2013-02-18 15:21:09.285317914 +0800
> +++ linux/mm/swapfile.c 2013-02-18 15:21:34.545004083 +0800
> @@ -2374,6 +2374,11 @@ static int __swap_duplicate(swp_entry_t
> goto unlock_out;
>
> count = p->swap_map[offset];
> + if (unlikely(swap_count(count) == SWAP_MAP_BAD)) {
> + err = -ENOENT;
> + goto unlock_out;
> + }
> +
> has_cache = count & SWAP_HAS_CACHE;
> count &= ~SWAP_HAS_CACHE;
> err = 0;
[PATCH] swap: fix swapoff ENOMEMs from discard
swapoff was sometimes failing with "Cannot allocate memory", coming
from try_to_unuse()'s -ENOMEM: it needs to allow for swap_duplicate()
failing on a free entry temporarily SWAP_MAP_BAD while being discarded.
We should use ACCESS_ONCE() there, and whenever accessing swap_map
locklessly; but rather than peppering it throughout try_to_unuse(),
just declare *swap_map with volatile.
try_to_unuse() is accustomed to *swap_map going down racily, but not
necessarily to it jumping up from 0 to SWAP_MAP_BAD: we'll be safer to
prevent that transition once SWP_WRITEOK is switched off, when it's a
waste of time to issue discards anyway (swapon can do a whole discard).
Signed-off-by: Hugh Dickins <hughd@xxxxxxxxxx>
---
mm/swapfile.c | 11 ++++++-----
1 file changed, 6 insertions(+), 5 deletions(-)
--- 3.9-rc1+shli/mm/swapfile.c 2013-02-24 12:14:25.799160751 -0800
+++ linux/mm/swapfile.c 2013-02-25 21:43:04.112050626 -0800
@@ -199,8 +199,8 @@ static void discard_swap_cluster(struct
static int swap_cluster_check_discard(struct swap_info_struct *si,
unsigned int idx)
{
-
- if (!(si->flags & SWP_DISCARDABLE))
+ if ((si->flags & (SWP_WRITEOK | SWP_DISCARDABLE)) !=
+ (SWP_WRITEOK | SWP_DISCARDABLE))
return 0;
/*
* If scan_swap_map() can't find a free cluster, it will check
@@ -1223,7 +1223,7 @@ static unsigned int find_next_to_unuse(s
else
continue;
}
- count = si->swap_map[i];
+ count = ACCESS_ONCE(si->swap_map[i]);
if (count && swap_count(count) != SWAP_MAP_BAD)
break;
}
@@ -1243,7 +1243,7 @@ int try_to_unuse(unsigned int type, bool
{
struct swap_info_struct *si = swap_info[type];
struct mm_struct *start_mm;
- unsigned char *swap_map;
+ volatile unsigned char *swap_map; /* ACCESS_ONCE throughout */
unsigned char swcount;
struct page *page;
swp_entry_t entry;
@@ -1294,7 +1294,8 @@ int try_to_unuse(unsigned int type, bool
* reused since sys_swapoff() already disabled
* allocation from here, or alloc_page() failed.
*/
- if (!*swap_map)
+ swcount = *swap_map;
+ if (!swcount || swcount == SWAP_MAP_BAD)
continue;
retval = -ENOMEM;
break;
--
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/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>