[IA64] remove dead BIO_VMERGE_BOUNDARY definition
The block layer dropped the virtual merge feature
(b8b3e16cfe
). BIO_VMERGE_BOUNDARY
definition is meaningless now (For IA64, BIO_VMERGE_BOUNDARY has been
meaningless for a long time since IA64 disables the virtual merge
feature).
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: Tony Luck <tony.luck@intel.com>
This commit is contained in:
parent
6a2d26fd3f
commit
d8d54b0252
|
@ -434,28 +434,4 @@ extern void memset_io(volatile void __iomem *s, int c, long n);
|
|||
|
||||
# endif /* __KERNEL__ */
|
||||
|
||||
/*
|
||||
* Enabling BIO_VMERGE_BOUNDARY forces us to turn off I/O MMU bypassing. It is said that
|
||||
* BIO-level virtual merging can give up to 4% performance boost (not verified for ia64).
|
||||
* On the other hand, we know that I/O MMU bypassing gives ~8% performance improvement on
|
||||
* SPECweb-like workloads on zx1-based machines. Thus, for now we favor I/O MMU bypassing
|
||||
* over BIO-level virtual merging.
|
||||
*/
|
||||
extern unsigned long ia64_max_iommu_merge_mask;
|
||||
#if 1
|
||||
#define BIO_VMERGE_BOUNDARY 0
|
||||
#else
|
||||
/*
|
||||
* It makes no sense at all to have this BIO_VMERGE_BOUNDARY macro here. Should be
|
||||
* replaced by dma_merge_mask() or something of that sort. Note: the only way
|
||||
* BIO_VMERGE_BOUNDARY is used is to mask off bits. Effectively, our definition gets
|
||||
* expanded into:
|
||||
*
|
||||
* addr & ((ia64_max_iommu_merge_mask + 1) - 1) == (addr & ia64_max_iommu_vmerge_mask)
|
||||
*
|
||||
* which is precisely what we want.
|
||||
*/
|
||||
#define BIO_VMERGE_BOUNDARY (ia64_max_iommu_merge_mask + 1)
|
||||
#endif
|
||||
|
||||
#endif /* _ASM_IA64_IO_H */
|
||||
|
|
Loading…
Reference in New Issue