Re: [PATCH/RFC] mm: don't cap request size based on read-ahead setting

From: Jens Axboe
Date: Wed Nov 16 2016 - 13:48:06 EST


On 11/16/2016 10:44 AM, Johannes Weiner wrote:
On Tue, Nov 15, 2016 at 03:41:58PM -0700, Jens Axboe wrote:
On 11/15/2016 03:27 PM, Johannes Weiner wrote:
Hi Jens,

On Thu, Nov 10, 2016 at 10:00:37AM -0700, Jens Axboe wrote:
Hi,

We ran into a funky issue, where someone doing 256K buffered reads saw
128K requests at the device level. Turns out it is read-ahead capping
the request size, since we use 128K as the default setting. This doesn't
make a lot of sense - if someone is issuing 256K reads, they should see
256K reads, regardless of the read-ahead setting.

To make matters more confusing, there's an odd interaction with the
fadvise hint setting. If we tell the kernel we're doing sequential IO on
this file descriptor, we can get twice the read-ahead size. But if we
tell the kernel that we are doing random IO, hence disabling read-ahead,
we do get nice 256K requests at the lower level. An application
developer will be, rightfully, scratching his head at this point,
wondering wtf is going on. A good one will dive into the kernel source,
and silently weep.

This patch introduces a bdi hint, io_pages. This is the soft max IO size
for the lower level, I've hooked it up to the bdev settings here.
Read-ahead is modified to issue the maximum of the user request size,
and the read-ahead max size, but capped to the max request size on the
device side. The latter is done to avoid reading ahead too much, if the
application asks for a huge read. With this patch, the kernel behaves
like the application expects.


diff --git a/block/blk-settings.c b/block/blk-settings.c
index f679ae122843..65f16cf4f850 100644
--- a/block/blk-settings.c
+++ b/block/blk-settings.c
@@ -249,6 +249,7 @@ void blk_queue_max_hw_sectors(struct request_queue *q,
unsigned int max_hw_secto
max_sectors = min_not_zero(max_hw_sectors, limits->max_dev_sectors);
max_sectors = min_t(unsigned int, max_sectors, BLK_DEF_MAX_SECTORS);
limits->max_sectors = max_sectors;
+ q->backing_dev_info.io_pages = max_sectors >> (PAGE_SHIFT - 9);

Could we simply set q->backing_dev_info.ra_pages here? This would
start the disk out with a less magical readahead setting than the
current 128k default, while retaining the ability for the user to
override it in sysfs later on. Plus, one less attribute to juggle.

We could, but then we'd have two places that tweak the same knob. I
think it's perfectly valid to have the read-ahead size be bigger than
the max request size, if you want some pipelining, for instance.

I'm not sure I follow. Which would be the two places and which knob?

It's actually three knobs, since I looked:

1) /sys/block/<dev>/queue/read_ahead_kb
2) /sys/class/bdi/<dev>/read_ahead_kb
3) /sys/block/<dev>/queue/max_sectors_kb

The first 2 control the same thing, bdi->ra_pages. If we update ->pages
from max_sectors_kb as well, that'd make three. At least the first two
tell you what they do, don't like having the max_sectors_kb fiddle it
automatically too.

What I meant how it could work is this: when the queue gets allocated,
we set ra_pages to the hard-coded 128K, like we do right now. When the
driver initializes and calls blk_queue_max_hw_sectors() it would set
ra_pages to the more informed, device-optimized max_sectors >>
(PAGE_SHIFT - 9). And once it's all initialized, the user can still
make adjustments to the default we picked in the kernel heuristic.

The 128k default is silly, though, that should be smarter. It should
probably default to the max request size.

Could you clarify the difference between max request size and what
blk_queue_max_hw_sectors() sets? The way I understood your patch is
that we want to use a readahead cap that's better suited to the
underlying IO device than the magic 128K. What am I missing?

This ties in with the above, so I'm just replying here. max_hw_sectors
is a hardware limit, the max size for a single command. max_sectors is a
user setting. For latency reasons, we don't necessarily want to issue
IOs up to the full hardware size, so we cap the kernel initiated size.
Eg for most hardware that can do 32-64MB in a command, we still limit
the default size to 512k.

The limit I added limits us to the user setting, not the hardware
setting.

@@ -369,10 +369,18 @@ ondemand_readahead(struct address_space *mapping,
bool hit_readahead_marker, pgoff_t offset,
unsigned long req_size)
{
- unsigned long max = ra->ra_pages;
+ unsigned long max_pages;
pgoff_t prev_offset;

/*
+ * Use the max of the read-ahead pages setting and the requested IO
+ * size, and then the min of that and the soft IO size for the
+ * underlying device.
+ */
+ max_pages = max_t(unsigned long, ra->ra_pages, req_size);
+ max_pages = min_not_zero(inode_to_bdi(mapping->host)->io_pages, max_pages);

This code would then go away, and it would apply the benefit of this
patch automatically to explicit readahead(2) and FADV_WILLNEED calls
going through force_page_cache_readahead() as well.

The path from the force actually works, which is why you get the weird
behavior with a file marked as RANDOM getting the full request size, and
not being limited by ra_pages.

How so? do_generic_file_read() calls page_cache_sync_readahead(), and
if the file is marked random it goes to force_page_cache_readahead():

void page_cache_sync_readahead(struct address_space *mapping,
struct file_ra_state *ra, struct file *filp,
pgoff_t offset, unsigned long req_size)
{
/* no read-ahead */
if (!ra->ra_pages)
return;

/* be dumb */
if (filp && (filp->f_mode & FMODE_RANDOM)) {
force_page_cache_readahead(mapping, filp, offset, req_size);
return;
}

/* do read-ahead */
ondemand_readahead(mapping, ra, filp, false, offset, req_size);
}

That function in turn still caps the reads to the default 128K ra_pages:

int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
pgoff_t offset, unsigned long nr_to_read)
{
if (unlikely(!mapping->a_ops->readpage && !mapping->a_ops->readpages))
return -EINVAL;

nr_to_read = min(nr_to_read, inode_to_bdi(mapping->host)->ra_pages);
while (nr_to_read) {
int err;

unsigned long this_chunk = (2 * 1024 * 1024) / PAGE_SIZE;

if (this_chunk > nr_to_read)
this_chunk = nr_to_read;
err = __do_page_cache_readahead(mapping, filp,
offset, this_chunk, 0);
if (err < 0)
return err;

offset += this_chunk;
nr_to_read -= this_chunk;
}
return 0;
}

How could you get IO requests bigger than the 128k ra_pages there?

Sorry, what I meant is that if we mark it random through fadvise, then
read-ahead doesn't get in the way and we get the full size. If we end up
through read-ahead, we get limited. So you are right, I should fix up
force_page_cache_readahead as well.

I'll send out a v3.

--
Jens Axboe