- Oct 19, 2007
-
-
Ingo Molnar authored
While we were reviewing pageattr_32/64.c for unification, Thomas Gleixner noticed the following serious SMP bug in global_flush_tlb(): down_read(&init_mm.mmap_sem); list_replace_init(&deferred_pages, &l); up_read(&init_mm.mmap_sem); this is SMP-unsafe because list_replace_init() done on two CPUs in parallel can corrupt the list. This bug has been introduced about a year ago in the 64-bit tree: commit ea7322de Author: Andi Kleen <ak@suse.de> Date: Thu Dec 7 02:14:05 2006 +0100 [PATCH] x86-64: Speed and clean up cache flushing in change_page_attr down_read(&init_mm.mmap_sem); - dpage = xchg(&deferred_pages, NULL); + list_replace_init(&deferred_pages, &l); up_read(&init_mm.mmap_sem); the xchg() based version was SMP-safe, but list_replace_init() is not. So this "cleanup" introduced a nasty bug. why this bug never become prominent is a mystery - it can probably be explained with the (still) relative obscurity of the x86_64 architecture. the safe fix for now is to write-lock init_mm.mmap_sem. Signed-off-by:
Ingo Molnar <mingo@elte.hu> Signed-off-by:
Thomas Gleixner <tglx@linutronix.de>
-
Linus Torvalds authored
* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6: pcnet32: remove private net_device_stats structure vortex_up should initialize "err" pcnet32: remove compile warnings in non-napi mode pcnet32: fix non-napi packet reception fix EMAC driver for proper napi_synchronize API sky2: shutdown cleanup napi_synchronize: waiting for NAPI forcedeth msi bugfix gianfar: fix obviously wrong #ifdef CONFIG_GFAR_NAPI placement fs_enet: Update for API changes gianfar: remove orphan struct. forcedeth: fix rx-work condition in nv_rx_process_optimized() too
-
Linus Torvalds authored
* master.kernel.org:/pub/scm/linux/kernel/git/bart/ide-2.6: (37 commits) ide: set drive->autotune in ide_pci_setup_ports() triflex: always tune PIO opti621: always tune PIO cy82c693: always tune PIO cs5520: always tune PIO alim15x3: always tune PIO ide: add IDE_HFLAG_LEGACY_IRQS host flag ide: add IDE_HFLAG_SERIALIZE host flag ide: add IDE_HFLAG_ERROR_STOPS_FIFO host flag piix: add DECLARE_ICH_DEV() macro pdc202xx_old: add DECLARE_PDC2026X_DEV() macro pdc202xx_new: add DECLARE_PDCNEW_DEV() macro aec62xx: no need to disable UDMA in ->init_hwif method for ATP850UF ide: remove .init_setup from ide_pci_device_t serverworks: remove ->init_setup scc_pata: remove ->init_setup pdc202xx_old: remove ->init_setup pdc202xx_new: remove ->init_setup hpt366: remove ->init_setup cmd64x: remove ->init_setup ...
-
Bartlomiej Zolnierkiewicz authored
Majority of host drivers using IDE PCI layer set drive->autotune, the only exceptions are: generic.c ns87415.c rz1000.c trm290.c * no ->set_pio_mode method it821x.c: * if memory allocation fails drive->autotune won't be set (but there also won't be ->set_pio_mode method in such case) piix.c: * MPIIX controller (no ->init_hwif method so also no ->set_pio_mode method) However if there is no ->set_pio_mode method there are no changes in behavior w.r.t. PIO tuning so always set drive->autotune in ide_pci_setup_ports(). Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Since cs5520 uses VDMA best PIO mode was tuned anyway by ide_dma_check() but only if DMA was successfully initialized. Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Add IDE_HFLAG_LEGACY_IRQS host flag to tell ide_pci_setup_ports() to set hwif->irq to legacy IRQ 14/15 (iff hwif->irq is not already set) and convert atiixp, piix, serverworks, sis5513 and slc90e66 host drivers to use it. While at it: * In piix.c add IDE_HFLAGS_PIIX define and don't use ->init_hwif for MPIIX. Acked-by:
Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Add IDE_HFLAG_SERIALIZE host flag to tell ide_pci_setup_ports() to set hwif/mate->serialized and convert aec62xx, cs5530 and sc1200 host drivers to use it. Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Add IDE_HFLAG_ERROR_STOPS_FIFO host flag and use it instead of hwif->err_stops_fifo. As a side-effect this change fixes hwif->err_stops_fifo not being restored by ide_hwif_restore(). Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Add DECLARE_ICH_DEV() macro. While at it: * Add init_hwif_ich() (->init_hwif method) for ICH controllers. * Rename init_chipset_piix() to init_chipset_ich() and use it only for ICH controllers. * Remove no longer needed piix_is_ichx() helper. Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
* No need to disable UDMA in ->init_hwif method for ATP850UF (and since we now always tune PIO it will be disabled by ->set_pio_mode calls anyway). * Bump driver version. Acked-by:
Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Now that all users were fixed we can safely remove it. Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Merge init_setup_{svwks,csb6}() into svwks_init_one(). While at it: * Remove redundant dev->device checks. * Operate on a local copy of serverworks_chipsets[] entry. * Use pci_resource_start(). Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Acked-by:
Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
* Split off pdc202ata4_fixup_irq() helper from init_setup_pdc202ata4(). * Merge init_setup_{pdc202ata4,pdc20265,pdc202xx}() into pdc202xx_init_one(). Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
* Split off pdc20270_get_dev2() helper from init_setup_pdc20270(). * Merge init_setup_{pdcnew,pdc20270,pdc20276}() into pdc202new_init_one(). While at it: * Change KERN_ level of interrupt fixup message from KERN_WARNING to KERN_INFO. Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
* Split off hpt{374,371,366}_init() helper from init_setup_hpt{374,371,366}(). * Merge init_setup_{374,372n,371,372a,302,366}() into hpt366_init_one(). While at it: * Use "HPT36x" name for HPT366/HPT368 chipsets. * Add .chip_name to struct hpt_info and use it to set set d->name. * Convert .max_ultra in struct hpt_info to .udma_mask and use it to set d->udma_mask. * Fix hpt302 to use HPT302_ALLOW_ATA133_6 define. * Change HPT366/HPT374 interrupt fixup message from KERN_WARNING to KERN_INFO. * Use the second hpt366_chipsets[] entry for HPT37x chipsets using HPT36x PCI device ID and fix .enablebits/.host_flags for HPT36x hpt366_chipsets[] entry. * Bump driver version. Cc: Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Merge init_setup_{cmd64x,cmd646}() into cmd64x_init_one(). Cc: Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Merge init_setup_{aec62xx,aec6x80}() into aec62xx_init_one(). While at it: * Use id->driver_data instead of dev->device. * Use ATA_UDMA6 define. Acked-by:
Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
v2: - bump host driver versions (as suggested by Sergei) - use I/O ops directly in drivers/ide/setup-pci.c Cc: Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Some places were using PCI_CLASS_REVISION instead of PCI_REVISION_ID so they were not converted by commit 44c10138 . Cc: Auke Kok <auke-jan.h.kok@intel.com> Cc: Greg Kroah-Hartman <gregkh@suse.de> Acked-by:
Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Auke Kok authored
Signed-off-by:
Auke Kok <auke-jan.h.kok@intel.com> Cc: Greg Kroah-Hartman <gregkh@suse.de> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Auke Kok authored
Signed-off-by:
Auke Kok <auke-jan.h.kok@intel.com> Cc: Greg Kroah-Hartman <gregkh@suse.de> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
* Add ->mwdma_mask and ->swdma_mask to ide_pci_device_t. * Set ide_hwif_t DMA masks using DMA masks from ide_pci_device_t in setup-pci.c::ide_pci_setup_ports() (iff DMA base is valid and ->init_hwif method may still override them). * Convert IDE PCI host drivers to use ide_pci_device_t DMA masks. While at it: * Use ATA_{UDMA,MWDMA,SWDMA}* defines. * hpt34x.c: add separate ide_pci_device_t instances for HPT343 and HPT345. * serverworks.c: fix DMA masks being set before checking DMA base. v2: * Add missing masks to DECLARE_GENERIC_PCI_DEV() macro. Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Documentation doesn't mention SWDMA and moreover all timings used for SWDMA modes were over-clocked when compared to ATA spec. Acked-by:
Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Add IDE_HFLAG_NO_LBA48[_DMA] host flags, use it instead of hwif->no_lba48[_dma] and then remove no longer needed hwif->no_lba48[_dma]. As a side-effect this change fixes hwif->no_lba48_dma not being restored by ide_hwif_restore(). Acked-by:
Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
* Make ide_pci_device_t.host_flags u32 and add IDE_HFLAG_CS5520 host flag. * Pass ide_pci_device_t *d to setup-pci.c::ide_get_or_set_dma_base() and use d->name instead of hwif->cds->name. * Set IDE_HFLAG_CS5520 host flag in cs5520 host driver and use it in ide_get_or_set_dma_base() to find out which PCI BAR to use, remove no longer needed cs5520.c::cs5520_init_setup_dma() and ide_pci_device_t.init_setup_dma. This fixes PCI bus-mastering not being checked for CS5510/CS5520 hosts. v2: * It is wrong to check simplex bits on CS5510/CS5520 as v1 did. (Noticed by Alan). Cc: Alan Cox <alan@lxorguk.ukuu.org.uk> Cc: Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Add IDE_HFLAG_NO_{DMA,AUTODMA} host flags. Convert all host drivers using ide_pci_device_t to use these flags instead of d->autodma and then remove no longer needed d->autodma. Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Add IDE_HFLAG_BOOTABLE host flag and IDE_HFLAG_OFF_BOARD define. Convert all host drivers using ide_pci_device_t to use IDE_HFLAG_{BOOTABLE,OFF_BOARD} instead of d->bootable and then remove no longer needed d->bootable. Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Add IDE_HFLAG_NO_ATAPI_DMA host flag and set it in host drivers which don't support ATAPI DMA. Then remove no longer needed hwif->atapi_dma. Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
* Add DECLARE_GENERIC_PCI_DEV() macro to generic IDE PCI host driver and use it to cleanup generic_chipsets[]. * Remove dead/obsolete code while at it. Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Bartlomiej Zolnierkiewicz authored
Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Benjamin Herrenschmidt authored
The cs5535 uses an incorrect construct to access the other drive of a pair, causing it to access beyond an array boundary on the secondary interface. This fixes it by using the new ide_get_paired_drive() helper instead. Bart: patch description fixes Signed-off-by:
Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Andrew Morton <akpm@osdl.org> Acked-by:
Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Benjamin Herrenschmidt authored
The siimage uses an incorrect construct to access the other drive of a pair, causing it to access beyond an array boundary on the secondary interface. This fixes it by using the new ide_get_paired_drive() helper instead. Bart: patch description fixes Signed-off-by:
Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Andrew Morton <akpm@osdl.org> Acked-by:
Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-
Benjamin Herrenschmidt authored
This adds a helper to get to the "other" drive on a pair connected to a given hwif. Signed-off-by:
Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Andrew Morton <akpm@osdl.org> Acked-by:
Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
-