diff options
| author | Linus Torvalds <torvalds@linux-foundation.org> | 2012-05-08 11:06:07 -0700 |
|---|---|---|
| committer | Linus Torvalds <torvalds@linux-foundation.org> | 2012-05-08 11:06:07 -0700 |
| commit | e9b19cd43f3f55a7f1d1c9a033feda8dabce3be6 (patch) | |
| tree | 2cb60b7af0bba9f69eb54db66f6e25f7c0fc898c /arch/x86 | |
| parent | 301cdf5c75695addaaf3b4857b6df7a1d764503e (diff) | |
| parent | d5e28005a1d2e67833852f4c9ea8ec206ea3ff85 (diff) | |
| download | op-kernel-dev-e9b19cd43f3f55a7f1d1c9a033feda8dabce3be6.zip op-kernel-dev-e9b19cd43f3f55a7f1d1c9a033feda8dabce3be6.tar.gz | |
Merge branch 'for-3.4-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu
Pull two percpu fixes from Tejun Heo:
"One adds missing KERN_CONT on split printk()s and the other makes
the percpu allocator avoid using PMD_SIZE as atom_size on x86_32.
Using PMD_SIZE led to vmalloc area exhaustion on certain
configurations (x86_32 android) and the only cost of using PAGE_SIZE
instead is static percpu area not being aligned to large page
mapping."
* 'for-3.4-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu:
percpu, x86: don't use PMD_SIZE as embedded atom_size on 32bit
percpu: use KERN_CONT in pcpu_dump_alloc_info()
Diffstat (limited to 'arch/x86')
| -rw-r--r-- | arch/x86/kernel/setup_percpu.c | 14 |
1 files changed, 13 insertions, 1 deletions
diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index 71f4727..5a98aa2 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -185,10 +185,22 @@ void __init setup_per_cpu_areas(void) #endif rc = -EINVAL; if (pcpu_chosen_fc != PCPU_FC_PAGE) { - const size_t atom_size = cpu_has_pse ? PMD_SIZE : PAGE_SIZE; const size_t dyn_size = PERCPU_MODULE_RESERVE + PERCPU_DYNAMIC_RESERVE - PERCPU_FIRST_CHUNK_RESERVE; + size_t atom_size; + /* + * On 64bit, use PMD_SIZE for atom_size so that embedded + * percpu areas are aligned to PMD. This, in the future, + * can also allow using PMD mappings in vmalloc area. Use + * PAGE_SIZE on 32bit as vmalloc space is highly contended + * and large vmalloc area allocs can easily fail. + */ +#ifdef CONFIG_X86_64 + atom_size = PMD_SIZE; +#else + atom_size = PAGE_SIZE; +#endif rc = pcpu_embed_first_chunk(PERCPU_FIRST_CHUNK_RESERVE, dyn_size, atom_size, pcpu_cpu_distance, |
