Searched refs:prefetches (Results 1 - 32 of 32) sorted by relevance

/linux-4.1.27/include/linux/
H A Dprefetch.h28 prefetch(x) - prefetches the cacheline at "x" for read
29 prefetchw(x) - prefetches the cacheline at "x" for write
30 spin_lock_prefetch(x) - prefetches the spinlock *x for taking
/linux-4.1.27/arch/sh/lib64/
H A Dcopy_page.S22 prefetches for the same cache set, so it's better to have the numbers
34 Do prefetches 4 lines ahead.
65 because they overlap with the time spent waiting for prefetches to
H A Dcopy_user_memcpy.S50 * erratum. The first two prefetches are nop-ed out to avoid upsetting the
/linux-4.1.27/arch/alpha/lib/
H A Dev6-copy_page.S39 The solution is to schedule the prefetches to avoid the memory
40 conflicts. I schedule the wh64 prefetches farther ahead of the
41 read prefetches to avoid this problem.
52 further by unrolling the loop and doing multiple prefetches per cycle.
/linux-4.1.27/drivers/md/persistent-data/
H A Ddm-transaction-manager.c98 struct prefetch_set prefetches; member in struct:dm_transaction_manager
179 prefetch_init(&tm->prefetches); dm_tm_create()
336 prefetch_add(&tm->real->prefetches, b); dm_tm_read_lock()
389 prefetch_issue(&tm->prefetches, tm->bm); dm_tm_issue_prefetches()
/linux-4.1.27/arch/sparc/include/asm/
H A Dprocessor_64.h232 * prefetches into the prefetch-cache which only is accessible prefetch()
234 * By contrast, "#one_write" prefetches into the L2 cache prefetch()
/linux-4.1.27/drivers/md/
H A Ddm-thin-metadata.h215 * Issue any prefetches that may be useful.
H A Ddm-verity.c544 * fills them. Then it issues prefetches and the I/O.
/linux-4.1.27/arch/mips/include/asm/octeon/
H A Dcvmx-asm.h113 /* complete prefetches, invalidate entire dcache */
/linux-4.1.27/drivers/crypto/
H A Dpadlock-aes.c211 * Padlock prefetches extra data so we must provide mapped input buffers. ecb_crypt_copy()
225 * Padlock prefetches extra data so we must provide mapped input buffers. cbc_crypt_copy()
/linux-4.1.27/arch/powerpc/platforms/cell/
H A Dras.c138 * page, in order to avoid prefetches in memcpy and similar cbe_ptcal_enable_on_node()
/linux-4.1.27/arch/ia64/lib/
H A Dcopy_page_mck.S31 * the prefetches. The four relevant points in the pipelined are called A, B, C, D:
/linux-4.1.27/arch/x86/kernel/
H A Damd_gart_64.c813 * automatic prefetches from the CPU allocating cache lines in gart_iommu_init()
841 * Any prefetches that hit unmapped entries won't get an bus abort gart_iommu_init()
/linux-4.1.27/arch/tile/lib/
H A Dmemcpy_32.S246 * prefetch that line more than once, or subsequent prefetches
406 /* Kick off two prefetches, but don't go past the end. */
/linux-4.1.27/arch/powerpc/perf/
H A Dpower8-pmu.c37 /* L1 cache data prefetches */
55 /* Total HW L3 prefetches(Load+store) */
/linux-4.1.27/arch/powerpc/platforms/powermac/
H A Dsleep.S178 /* Flush any pending L2 data prefetches to work around HW bug */
/linux-4.1.27/arch/sparc/lib/
H A DU3memcpy.S69 * 3) This code never prefetches cachelines past the end
/linux-4.1.27/arch/sparc/mm/
H A Dfault_64.c372 * have to avoid prefetches which also have bit 21 set. do_sparc64_fault()
/linux-4.1.27/arch/mips/mm/
H A Duasm.c346 * As per erratum Core-14449, replace prefetches 0-4, I_u3u1u2()
/linux-4.1.27/drivers/infiniband/hw/mlx5/
H A Dodp.c621 * prefetches more pages. The second operation cannot use the pfault mlx5_ib_mr_rdma_pfault_handler()
/linux-4.1.27/arch/arm/mm/
H A Ddma-mapping.c51 * speculative prefetches. We model our approach on the assumption that
52 * the CPU does do speculative prefetches, which means we clean caches
/linux-4.1.27/arch/x86/include/asm/
H A Dprocessor.h804 * It's not worth to care about 3dnow prefetches for the K6
/linux-4.1.27/arch/mips/cavium-octeon/executive/
H A Dcvmx-l2c.c646 * Make sure core is quiet (no prefetches, etc.) before __read_l2_tag()
/linux-4.1.27/arch/parisc/kernel/
H A Dperf_images.h1570 * or not, but does *not* include I-cache prefetches, which are generated by
1579 * should be between 1 and 2. If it is close to 1, most prefetches are
1580 * eventually called for by the IFU; if it is close to 2, almost no prefetches
3093 * prefetches).
/linux-4.1.27/arch/sh/kernel/cpu/sh5/
H A Dentry.S960 /* Do prefetches */
/linux-4.1.27/arch/ia64/hp/common/
H A Dsba_iommu.c80 ** If a device prefetches beyond the end of a valid pdir entry, it will cause
/linux-4.1.27/tools/perf/util/
H A Devsel.c385 { "prefetch", "prefetches", "speculative-read", "speculative-load", },
/linux-4.1.27/drivers/net/wireless/ath/ath9k/
H A Dhw.c1163 * set AHB_MODE not to do cacheline prefetches ath9k_hw_set_dma()
/linux-4.1.27/drivers/gpu/drm/i915/
H A Di915_gem_gtt.c2022 * There are a number of places where the hardware apparently prefetches i915_gem_setup_global_gtt()
/linux-4.1.27/arch/x86/kernel/cpu/
H A Dperf_event_intel.c442 * - prefetches are not included in the counts because they are not
/linux-4.1.27/drivers/scsi/
H A Dncr53c8xx.c1787 ** between CPU and 53c720 does prefetches, which causes
3400 ** between CPU and 53c720 does prefetches, which causes
/linux-4.1.27/drivers/net/ethernet/chelsio/cxgb3/
H A Dsge.c1827 * receive handler. Batches need to be of modest size as we do prefetches

Completed in 1301 milliseconds