[dm-devel] [PATCH 0/4] Fix a crash when block device is read and block size is changed at the same time

[Jeff Moyer]

Mikulas Patocka <mpatocka at redhat.com> writes:

> On Tue, 18 Sep 2012, Jeff Moyer wrote:
>
>> Mikulas Patocka <mpatocka at redhat.com> writes:
>> 
>> > Hi Jeff
>> >
>> > Thanks for testing.
>> >
>> > It would be interesting ... what happens if you take the patch 3, leave 
>> > "struct percpu_rw_semaphore bd_block_size_semaphore" in "struct 
>> > block_device", but remove any use of the semaphore from fs/block_dev.c? - 
>> > will the performance be like unpatched kernel or like patch 3? It could be 
>> > that the change in the alignment affects performance on your CPU too, just 
>> > differently than on my CPU.
>> 
>> I'll give it a try and report back.
>> 
>> > What is the CPU model that you used for testing?
>> 
>> http://ark.intel.com/products/53570/Intel-Xeon-Processor-E7-2860-%2824M-Cache-2_26-GHz-6_40-GTs-Intel-QPI%29
>> 
> BTW. why did you use just 4 processes? - the processor has 10 cores and 20 
> threads (so theoretically, you could run 20 processes bound on a single 
> numa node). Were the results not stable with more than 4 processes?

There is no good reason for it.  Since I was able to show some
differences in performance, I didn't see the need to scale beyond 4.  I
can certainly bump the count up if/when that becomes interesting.

Cheers,
Jeff