summaryrefslogtreecommitdiffstats
path: root/drivers
diff options
context:
space:
mode:
authorNick Piggin <npiggin@suse.de>2007-10-16 01:24:40 -0700
committerLinus Torvalds <torvalds@woody.linux-foundation.org>2007-10-16 09:42:53 -0700
commit557ed1fa2620dc119adb86b34c614e152a629a80 (patch)
treed00b31a7f197583c2bd8fffa1fd135fbbb5d6abc /drivers
parentaadb4bc4a1f9108c1d0fbd121827c936c2ed4217 (diff)
downloadop-kernel-dev-557ed1fa2620dc119adb86b34c614e152a629a80.zip
op-kernel-dev-557ed1fa2620dc119adb86b34c614e152a629a80.tar.gz
remove ZERO_PAGE
The commit b5810039a54e5babf428e9a1e89fc1940fabff11 contains the note A last caveat: the ZERO_PAGE is now refcounted and managed with rmap (and thus mapcounted and count towards shared rss). These writes to the struct page could cause excessive cacheline bouncing on big systems. There are a number of ways this could be addressed if it is an issue. And indeed this cacheline bouncing has shown up on large SGI systems. There was a situation where an Altix system was essentially livelocked tearing down ZERO_PAGE pagetables when an HPC app aborted during startup. This situation can be avoided in userspace, but it does highlight the potential scalability problem with refcounting ZERO_PAGE, and corner cases where it can really hurt (we don't want the system to livelock!). There are several broad ways to fix this problem: 1. add back some special casing to avoid refcounting ZERO_PAGE 2. per-node or per-cpu ZERO_PAGES 3. remove the ZERO_PAGE completely I will argue for 3. The others should also fix the problem, but they result in more complex code than does 3, with little or no real benefit that I can see. Why? Inserting a ZERO_PAGE for anonymous read faults appears to be a false optimisation: if an application is performance critical, it would not be doing many read faults of new memory, or at least it could be expected to write to that memory soon afterwards. If cache or memory use is critical, it should not be working with a significant number of ZERO_PAGEs anyway (a more compact representation of zeroes should be used). As a sanity check -- mesuring on my desktop system, there are never many mappings to the ZERO_PAGE (eg. 2 or 3), thus memory usage here should not increase much without it. When running a make -j4 kernel compile on my dual core system, there are about 1,000 mappings to the ZERO_PAGE created per second, but about 1,000 ZERO_PAGE COW faults per second (less than 1 ZERO_PAGE mapping per second is torn down without being COWed). So removing ZERO_PAGE will save 1,000 page faults per second when running kbuild, while keeping it only saves less than 1 page clearing operation per second. 1 page clear is cheaper than a thousand faults, presumably, so there isn't an obvious loss. Neither the logical argument nor these basic tests give a guarantee of no regressions. However, this is a reasonable opportunity to try to remove the ZERO_PAGE from the pagefault path. If it is found to cause regressions, we can reintroduce it and just avoid refcounting it. The /dev/zero ZERO_PAGE usage and TLB tricks also get nuked. I don't see much use to them except on benchmarks. All other users of ZERO_PAGE are converted just to use ZERO_PAGE(0) for simplicity. We can look at replacing them all and maybe ripping out ZERO_PAGE completely when we are more satisfied with this solution. Signed-off-by: Nick Piggin <npiggin@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus "snif" Torvalds <torvalds@linux-foundation.org>
Diffstat (limited to 'drivers')
-rw-r--r--drivers/char/mem.c125
1 files changed, 17 insertions, 108 deletions
diff --git a/drivers/char/mem.c b/drivers/char/mem.c
index bbee97f..64551ab 100644
--- a/drivers/char/mem.c
+++ b/drivers/char/mem.c
@@ -625,65 +625,10 @@ static ssize_t splice_write_null(struct pipe_inode_info *pipe,struct file *out,
return splice_from_pipe(pipe, out, ppos, len, flags, pipe_to_null);
}
-#ifdef CONFIG_MMU
-/*
- * For fun, we are using the MMU for this.
- */
-static inline size_t read_zero_pagealigned(char __user * buf, size_t size)
-{
- struct mm_struct *mm;
- struct vm_area_struct * vma;
- unsigned long addr=(unsigned long)buf;
-
- mm = current->mm;
- /* Oops, this was forgotten before. -ben */
- down_read(&mm->mmap_sem);
-
- /* For private mappings, just map in zero pages. */
- for (vma = find_vma(mm, addr); vma; vma = vma->vm_next) {
- unsigned long count;
-
- if (vma->vm_start > addr || (vma->vm_flags & VM_WRITE) == 0)
- goto out_up;
- if (vma->vm_flags & (VM_SHARED | VM_HUGETLB))
- break;
- count = vma->vm_end - addr;
- if (count > size)
- count = size;
-
- zap_page_range(vma, addr, count, NULL);
- if (zeromap_page_range(vma, addr, count, PAGE_COPY))
- break;
-
- size -= count;
- buf += count;
- addr += count;
- if (size == 0)
- goto out_up;
- }
-
- up_read(&mm->mmap_sem);
-
- /* The shared case is hard. Let's do the conventional zeroing. */
- do {
- unsigned long unwritten = clear_user(buf, PAGE_SIZE);
- if (unwritten)
- return size + unwritten - PAGE_SIZE;
- cond_resched();
- buf += PAGE_SIZE;
- size -= PAGE_SIZE;
- } while (size);
-
- return size;
-out_up:
- up_read(&mm->mmap_sem);
- return size;
-}
-
static ssize_t read_zero(struct file * file, char __user * buf,
size_t count, loff_t *ppos)
{
- unsigned long left, unwritten, written = 0;
+ size_t written;
if (!count)
return 0;
@@ -691,69 +636,33 @@ static ssize_t read_zero(struct file * file, char __user * buf,
if (!access_ok(VERIFY_WRITE, buf, count))
return -EFAULT;
- left = count;
-
- /* do we want to be clever? Arbitrary cut-off */
- if (count >= PAGE_SIZE*4) {
- unsigned long partial;
+ written = 0;
+ while (count) {
+ unsigned long unwritten;
+ size_t chunk = count;
- /* How much left of the page? */
- partial = (PAGE_SIZE-1) & -(unsigned long) buf;
- unwritten = clear_user(buf, partial);
- written = partial - unwritten;
- if (unwritten)
- goto out;
- left -= partial;
- buf += partial;
- unwritten = read_zero_pagealigned(buf, left & PAGE_MASK);
- written += (left & PAGE_MASK) - unwritten;
+ if (chunk > PAGE_SIZE)
+ chunk = PAGE_SIZE; /* Just for latency reasons */
+ unwritten = clear_user(buf, chunk);
+ written += chunk - unwritten;
if (unwritten)
- goto out;
- buf += left & PAGE_MASK;
- left &= ~PAGE_MASK;
- }
- unwritten = clear_user(buf, left);
- written += left - unwritten;
-out:
- return written ? written : -EFAULT;
-}
-
-static int mmap_zero(struct file * file, struct vm_area_struct * vma)
-{
- int err;
-
- if (vma->vm_flags & VM_SHARED)
- return shmem_zero_setup(vma);
- err = zeromap_page_range(vma, vma->vm_start,
- vma->vm_end - vma->vm_start, vma->vm_page_prot);
- BUG_ON(err == -EEXIST);
- return err;
-}
-#else /* CONFIG_MMU */
-static ssize_t read_zero(struct file * file, char * buf,
- size_t count, loff_t *ppos)
-{
- size_t todo = count;
-
- while (todo) {
- size_t chunk = todo;
-
- if (chunk > 4096)
- chunk = 4096; /* Just for latency reasons */
- if (clear_user(buf, chunk))
- return -EFAULT;
+ break;
buf += chunk;
- todo -= chunk;
+ count -= chunk;
cond_resched();
}
- return count;
+ return written ? written : -EFAULT;
}
static int mmap_zero(struct file * file, struct vm_area_struct * vma)
{
+#ifndef CONFIG_MMU
return -ENOSYS;
+#endif
+ if (vma->vm_flags & VM_SHARED)
+ return shmem_zero_setup(vma);
+ return 0;
}
-#endif /* CONFIG_MMU */
static ssize_t write_full(struct file * file, const char __user * buf,
size_t count, loff_t *ppos)
OpenPOWER on IntegriCloud