[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [dm-devel] [PATCH] fix default IO restrictions



Am Di, den 27.01.2004 schrieb Mike Christie um 20:36:

> The queue's max_sectors, max_[phys/hw]_segments, max_segments_size etc 
> are device driver/HW limitations. blk_init_queue does not know an 
> optimal value for every device, but it might be able to set a safe 
> default value. Maybe a better name for MAX_SECTORS is DEFAULT_MAX_SECTORS.
> 
> BIO_MAX_[SECTORS/SIZE/PAGES] is what you called a "system-wide limit". A 
> bio can only be so large (the bvec_pool is hardcoded). However when 
> merging the bio, q->max_sectors is one of the indicators that tell the 
> queue when a request is too large for the device.

Ah, yes. I see now. BIO_MAX_SECTORS is something like 2048 (the real
hard limit of the bio subsystem) on i386 while MAX_SECTORS is 255...
just a random sane value. Thanks.

So, why not removing set_default_limits and leaving the restriction
values at zero and add a function that replaces allo remaining
restrictions that are still zero with the default restrictions when the
table polulation completed after all targets are added?





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]