- 29 Apr, 2008 12 commits
-
-
Nishanth Aravamudan authored
Because of page order checks in __alloc_pages(), hugepage (and similarly large order) allocations will not retry unless explicitly marked __GFP_REPEAT. However, the current retry logic is nearly an infinite loop (or until reclaim does no progress whatsoever). For these costly allocations, that seems like overkill and could potentially never terminate. Mel observed that allowing current __GFP_REPEAT semantics for hugepage allocations essentially killed the system. I believe this is because we may continue to reclaim small orders of pages all over, but never have enough to satisfy the hugepage allocation request. This is clearly only a problem for large order allocations, of which hugepages are the most obvious (to me). Modify try_to_free_pages() to indicate how many pages were reclaimed. Use that information in __alloc_pages() to eventually fail a large __GFP_REPEAT allocation when we've reclaimed an order of pages equal to or greater than the allocation's order. This relies on lumpy reclaim functioning as advertised. Due to fragmentation, lumpy reclaim may not be able to free up the order needed in one invocation, so multiple iterations may be requred. In other words, the more fragmented memory is, the more retry attempts __GFP_REPEAT will make (particularly for higher order allocations). This changes the semantics of __GFP_REPEAT subtly, but *only* for allocations > PAGE_ALLOC_COSTLY_ORDER. With this patch, for those size allocations, we will try up to some point (at least 1<<order reclaimed pages), rather than forever (which is the case for allocations <= PAGE_ALLOC_COSTLY_ORDER). This change improves the /proc/sys/vm/nr_hugepages interface with a follow-on patch that makes pool allocations use __GFP_REPEAT. Rather than administrators repeatedly echo'ing a particular value into the sysctl, and forcing reclaim into action manually, this change allows for the sysctl to attempt a reasonable effort itself. Similarly, dynamic pool growth should be more successful under load, as lumpy reclaim can try to free up pages, rather than failing right away. Choosing to reclaim only up to the order of the requested allocation strikes a balance between not failing hugepage allocations and returning to the caller when it's unlikely to every succeed. Because of lumpy reclaim, if we have freed the order requested, hopefully it has been in big chunks and those chunks will allow our allocation to succeed. If that isn't the case after freeing up the current order, I don't think it is likely to succeed in the future, although it is possible given a particular fragmentation pattern. Signed-off-by:
Nishanth Aravamudan <nacc@us.ibm.com> Cc: Andy Whitcroft <apw@shadowen.org> Tested-by:
Mel Gorman <mel@csn.ul.ie> Cc: Dave Hansen <haveblue@us.ibm.com> Cc: Christoph Lameter <clameter@sgi.com> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
Nishanth Aravamudan authored
The definition and use of __GFP_REPEAT, __GFP_NOFAIL and __GFP_NORETRY in the core VM have somewhat differing comments as to their actual semantics. Annoyingly, the flags definition has inline and header comments, which might be interpreted as not being equivalent. Just add references to the header comments in the inline ones so they don't go out of sync in the future. In their use in __alloc_pages() clarify that the current implementation treats low-order allocations and __GFP_REPEAT allocations as distinct cases. To clarify, the flags' semantics are: __GFP_NORETRY means try no harder than one run through __alloc_pages __GFP_REPEAT means __GFP_NOFAIL __GFP_NOFAIL means repeat forever order <= PAGE_ALLOC_COSTLY_ORDER means __GFP_NOFAIL Signed-off-by:
-
KAMEZAWA Hiroyuki authored
usemap must be initialized only when pfn is within zone. If not, it corrupts memory. And this patch also reduces the number of calls to set_pageblock_migratetype() from (pfn & (pageblock_nr_pages -1) to !(pfn & (pageblock_nr_pages-1) it should be called once per pageblock. Signed-off-by:
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Acked-by:
-
Harvey Harrison authored
Signed-off-by:
Harvey Harrison <harvey.harrison@gmail.com> Signed-off-by:
-
Harvey Harrison authored
drivers/media/video/v4l2-common.c:719:16: warning: Using plain integer as NULL pointer drivers/media/video/au0828/au0828-dvb.c:122:19: warning: Using plain integer as NULL pointer drivers/media/video/ivtv/ivtv-yuv.c:1101:22: warning: Using plain integer as NULL pointer drivers/media/video/ivtv/ivtv-yuv.c:1102:23: warning: Using plain integer as NULL pointer drivers/media/video/pvrusb2/pvrusb2-audio.c:78:39: warning: Using plain integer as NULL pointer drivers/media/video/pvrusb2/pvrusb2-video-v4l.c:84:39: warning: Using plain integer as NULL pointer drivers/media/video/pvrusb2/pvrusb2-v4l2.c:1264:9: warning: Using plain integer as NULL pointer drivers/media/video/pvrusb2/pvrusb2-context.c:197:28: warning: Using plain integer as NULL pointer drivers/media/video/pvrusb2/pvrusb2-cx2584x-v4l.c:126:39: warning: Using plain integer as NULL pointer drivers/media/video/pvrusb2/pvrusb2-dvb.c:133:32: warning: Using plain integer as NULL pointer drivers/media/video/pvrusb2/pvrusb2-dvb.c:145:31: warning: Using plain integer as NULL pointer drivers/media/video/pvrusb2/pvrusb2-dvb.c:177:55: warning: Using plain integer as NULL pointer drivers/media/video/videobuf-core.c:100:9: warning: Using plain integer as NULL pointer Signed-off-by:
-
Harvey Harrison authored
drivers/scsi/aic7xxx/aic7770_osm.c:53:58: warning: Using plain integer as NULL pointer drivers/scsi/aic7xxx/aic7xxx_osm_pci.c:355:47: warning: Using plain integer as NULL pointer drivers/scsi/aic7xxx/aic7xxx_osm_pci.c:372:55: warning: Using plain integer as NULL pointer drivers/scsi/aha152x.c:997:28: warning: Using plain integer as NULL pointer drivers/scsi/aha152x.c:1003:28: warning: Using plain integer as NULL pointer drivers/scsi/aha152x.c:1165:46: warning: Using plain integer as NULL pointer drivers/scsi/fdomain.c:1446:40: warning: Using plain integer as NULL pointer drivers/scsi/sym53c8xx_2/sym_hipd.c:1650:51: warning: Using plain integer as NULL pointer drivers/scsi/sym53c8xx_2/sym_hipd.c:3171:42: warning: Using plain integer as NULL pointer drivers/scsi/sym53c8xx_2/sym_hipd.c:5732:52: warning: Using plain integer as NULL pointer drivers/scsi/ncr53c8xx.c:8189:31: warning: Using plain integer as NULL pointer drivers/scsi/ncr53c8xx.c:8225:34: warning: Using plain integer as NULL pointer drivers/scsi/dpt_i2o.c:156:32: warning: Using plain integer as NULL pointer drivers/scsi/ultrastor.c:954:42: warning: Using plain integer as NULL pointer drivers/scsi/ultrastor.c:1104:18: warning: Using plain integer as NULL pointer Signed-off-by:
-
git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6Linus Torvalds authored
* git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6: (35 commits) siimage: coding style cleanup (take 2) ide-cd: clean up cdrom_analyze_sense_data() ide-cd: fix test unsigned var < 0 ide: add TSSTcorp CDDVDW SH-S202H to ivb_list[] piix: add Asus Eee 701 controller to short cable list ARM: always select HAVE_IDE remove the broken ETRAX_IDE driver ide: remove ->dma_prdtable field from ide_hwif_t ide: remove ->dma_vendor{1,3} fields from ide_hwif_t scc_pata: add ->dma_host_set and ->dma_start methods ide: skip "VLB sync" if host uses MMIO ide: add ide_pad_transfer() helper ide: remove ->INW and ->OUTW methods ide: use IDE I/O helpers directly in ide_tf_{load,read}() ns87415: add ->tf_read method scc_pata: add ->tf_{load,read} methods ide-h8300: add ->tf_{load,read} methods ide-cris: add ->tf_{load,read} methods ide: add ->tf_load and ->tf_read methods ide: move ide_tf_{load,read} to ide-iops.c ...
-
git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuildLinus Torvalds authored
* git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild: kconfig: add named choice group kconfig: fix choice dependency check kconifg: 'select' considered less evil dontdiff: ignore timeconst.h dontdiff: add modules.order kbuild: fix unportability in gen_initramfs_list.sh kbuild: fix help output to show correct arch kbuild: show defconfig subdirs in make help kconfig: reversed borderlines in inputbox
-
Harvey Harrison authored
drivers/atm/nicstar.c:418:25: warning: Using plain integer as NULL pointer drivers/char/drm/r128_cce.c:820:25: warning: Using plain integer as NULL pointer drivers/char/tty_io.c:1183:10: warning: Using plain integer as NULL pointer Signed-off-by:
-
Harvey Harrison authored
mm/hugetlb.c:207:11: warning: Using plain integer as NULL pointer Signed-off-by:
-
Harvey Harrison authored
kernel/cpuset.c:1268:52: warning: Using plain integer as NULL pointer kernel/pid_namespace.c:95:24: warning: Using plain integer as NULL pointer Signed-off-by:
Paul Jackson <pj@sgi.com> Signed-off-by:
-
Harvey Harrison authored
init/do_mounts_rd.c:215:13: warning: Using plain integer as NULL pointer init/do_mounts_md.c:136:45: warning: Using plain integer as NULL pointer Signed-off-by:
-
- 28 Apr, 2008 28 commits
-
-
Sergei Shtylyov authored
Fix 18 errors and several warnings given by checkpatch.pl: - use of C99 // comments; - trailing whitespace; - 'switch' and 'case' not at the same indentation level; - no space before the open parenthesis of the 'if' and 'switch' statements; - space between function name and open parenthesis (though I have introduced such warnins in some places since the code looks prettier with the spaces); - including <asm/io.h> instead of <linux/io.h>; - line over 80 characters. In addition to these changes, also do the following: - make the arrays in sil_set_pio_mode() 'static', and make the arrays in sil_set_dma_mode() 'static const'; - change the string of the 'if' statements into the 'switch' statement in sil_pata_udma_filter(); - drop the needless '==' operators from the 'if' statements where a condition is a mere bit test; - remove needless initializer for the 'tmp' variable in init_chipset_siimage(); - beautify groups of the variable initializers and assignment operators; - add new line after variable definitions; - remove new line between the comment and the statements it refers to; - remove needless curly braces and parentheses; - fix typos, capitalize acronyms, etc. in the comments... Signed-off-by:
Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Roel Kluin authored
[bart: fix handling of bio_sectors(failed_command->bio) == 0] Signed-off-by:
Roel Kluin <12o3l@tiscali.nl> Cc: Borislav Petkov <petkovbb@gmail.com> Signed-off-by:
-
Roel Kluin authored
valid is unsigned and cannot be below 0. Signed-off-by:
-
Alexander Smal authored
Signed-off-by: -
Bartlomiej Zolnierkiewicz authored
Based on ata_piix patch by Dan McGee. Signed-off-by: -
Adrian Bunk authored
It's plain wrong for PCMCIA to select HAVE_IDE that implies e.g. the availability of an asm/ide.h It turns out this was done for ARM, and we can simply always select HAVE_IDE on ARM instead of manually tracking which platforms might possible have an IDE controller directly or indirectly. Signed-off-by:
Adrian Bunk <bunk@kernel.org> Cc: Russell King <rmk+lkml@arm.linux.org.uk> Cc: Sam Ravnborg <sam@ravnborg.org> Signed-off-by:
-
Adrian Bunk authored
ETRAX_IDE was marked as broken last year with the comment "it doesn't even compile currently". Remove it since it won't get fixed in the near future. On Mon, Apr 14, 2008 at 02:50:19PM +0200, Mikael Starvik wrote: > You can remove it for now and we will resubmit a new if/when we get around > to fix it. [bart: ported it over IDE tree] Signed-off-by:
-
Bartlomiej Zolnierkiewicz authored
* Use 'hwif->dma_base + {4,8}' instead of hwif->dma_prdtable in {ide,scc}_dma_setup(). * Remove no longer needed ->dma_prdtable field from ide_hwif_t. While at it: * Use ATA_DMA_TABLE_OFS define. Acked-by: -
Bartlomiej Zolnierkiewicz authored
* Use 'hwif->dma_base + {1,3}' instead of hwif->dma_vendor{1,3} in pdc202xx_new host driver. * Remove no longer needed ->dma_vendor{1,3} fields from ide_hwif_t. Acked-by: -
Bartlomiej Zolnierkiewicz authored
Add ->dma_host_set and ->dma_start methods (+ __scc_dma_end() helper) so scc_ide_{in,out}b() can be used directly. There should be no functional changes caused by this patch. Acked-by: -
Bartlomiej Zolnierkiewicz authored
* Skip "VLB sync" in ata_{in,out}put_data() if host uses MMIO. * Use I/O ops directly in ata_vlb_sync() an drop no longer needed 'ide_drive_t *drive' argument. Acked-by: -
Bartlomiej Zolnierkiewicz authored
* Add ide_pad_transfer() helper (which uses ->{in,out}put_data methods internally so the transfer is also padded to drive+host requirements) and use it instead of ide_atapi_{write_zeros,discard_data}(). * Remove no longer needed ide_atapi_{write_zeros,discard_data}(). Cc: Borislav Petkov <petkovbb@gmail.com> Acked-by: -
Bartlomiej Zolnierkiewicz authored
* Remove no longer used ->INW and ->OUTW methods. While at it: * scc_pata.c: scc_ide_{out,in}w() is called only in scc_tf_{load,read}() so inline it there. Signed-off-by: -
Bartlomiej Zolnierkiewicz authored
There should be no functional changes caused by this patch. Signed-off-by: -
Bartlomiej Zolnierkiewicz authored
Add ->tf_read method so out{b,w}(), in{b,w}() and superio_ide_inb() can be used directly. There should be no functional changes caused by this patch. Signed-off-by: -
Bartlomiej Zolnierkiewicz authored
Add ->tf_{load,read} methods so scc_ide_{outb,outw,inb,inw}() can be used directly. There should be no functional changes caused by this patch. Signed-off-by: -
Bartlomiej Zolnierkiewicz authored
Add ->tf_{load,read} methods so outb()/inb() and mm_outw()/mm_inw() can be used directly. There should be no functional changes caused by this patch. Signed-off-by: -
Bartlomiej Zolnierkiewicz authored
Add ->tf_{load,read} methods so cris_ide_{outb,outw,inb,inw}() can be used directly. There should be no functional changes caused by this patch. Signed-off-by: -
Bartlomiej Zolnierkiewicz authored
* Add ->tf_load and ->tf_read methods to ide_hwif_t and set the default methods in default_hwif_transport(). * Use ->tf_{load,read} instead o calling ide_tf_{load,read}() directly. * Make ide_tf_{load,read}() static. There should be no functional changes caused by this patch. Acked-by: -
Bartlomiej Zolnierkiewicz authored
Signed-off-by: -
Bartlomiej Zolnierkiewicz authored
Factor out debugging code from ide_tf_load() to ide_tf_dump() helper and update ide_tf_load() users accordingly. Signed-off-by: -
Bartlomiej Zolnierkiewicz authored
Add ide_execute_pkt_cmd() helper for executing PACKET command, then convert ATAPI device drivers to use it. As a nice side-effect this fixes ide-{floppy,tape,scsi} w.r.t. ide_lock taking (ide-cd was OK). Signed-off-by: -
Bartlomiej Zolnierkiewicz authored
Signed-off-by: -
Bartlomiej Zolnierkiewicz authored
Always use ->OUTBSYNC method for executing commands so the posting is done if needed (this affects only pmac and scc_pata host drivers at the moment). Signed-off-by: -
Bartlomiej Zolnierkiewicz authored
* proc_reports_siimage() is now only called by init_chipset_siimage() so inline it there. * Use array instead of switch statement for reporting clock modes. There should be no functional changes caused by this patch. Acked-by:
-
Bartlomiej Zolnierkiewicz authored
Add sil_iowrite{8,16,32}() and sil_ioread{8,16}() helpers, then use them to merge code accessing configuration registers through PCI and MMIO together. [ because of this SATA initialization bits from setup_mmio_siimage() are moved to init_chipset_siimage() ] This also cuts code size a bit: text data bss dec hex filename 4437 164 0 4601 11f9 drivers/ide/pci/siimage.o.before 3979 164 0 4143 102f drivers/ide/pci/siimage.o.after While at it: * Use I/O ops directly instead of using ->IN{B,W} and ->OUT{B,W}. * Fixup CodingStyle in setup_mmio_siimage(). * Rename 'tmpbyte' variable to 'tmp' in init_chipset_siimage(). There should be no functional changes caused by this patch. Acked-by: -
Bartlomiej Zolnierkiewicz authored
Do clocking register posting earlier in setup_mmio_siimage() to match code in init_chipset_siimage(). This is a preparation for the next patch which merges PCI and MMIO code paths together. Acked-by:
-
Bartlomiej Zolnierkiewicz authored
* Use ins{w,l}()/outs{w,l}() and __ide_mm_ins{w,l}()/__ide_mm_outs{w,l}() directly in ata_{in,out}put_data() (by using IDE_HFLAG_MMIO host flag to decide which I/O ops are required). * Remove no longer needed ->INS{W,L} and ->OUTS{W,L} methods (ide-h8300, au1xxx-ide and scc_pata implement their own ->{in,out}put_data methods). There should be no functional changes caused by this patch. Signed-off-by:
-
