summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* MIPS: Consolidate all CONFIG_CPU_HAS_LLSC use in a single C file.Ralf Baechle2009-09-174-9/+12
| | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Clean up linker script using new linker script macros.Nelson Elhage2009-09-171-78/+8
| | | | | | | | | This patch results in fewer output sections and in some data being reordered, but should have no functional impact. Signed-off-by: Nelson Elhage <nelhage@ksplice.com> Signed-off-by: Tim Abbott <tabbott@ksplice.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Use PAGE_SIZE in assembly instead of _PAGE_SIZE.Nelson Elhage2009-09-173-11/+10
| | | | | | | | | | | | Now that PAGE_SIZE is available to assembly directly, there is no need to separately expose it as _PAGE_SIZE through asm-offsets. In addition, remove _PAGE_SHIFT from asm-offsets, since it was never needed, and is not used anywhere. Signed-off-by: Nelson Elhage <nelhage@ksplice.com> Signed-off-by: Tim Abbott <tabbott@ksplice.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Convert to asm-generic/hardirq.hChristoph Hellwig2009-09-171-9/+3
| | | | | Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: AR7: Make board code register ar7_wdt as a platform deviceFlorian Fainelli2009-09-171-0/+32
| | | | | | | | | This patch makes the board code register the ar7_wdt driver as a platform device. We move the dynamic resource calculation here since the driver should not be aware of the AR7 SoC version it is running on. Signed-off-by: Florian Fainelli <florian@openwrt.org> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: Update the default config file for fuloong2eWu Zhangjin2009-09-171-283/+190
| | | | | Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: Add GCC 4.4 support for Loongson2EWu Zhangjin2009-09-173-9/+17
| | | | | | | | | | | | | | | | Because only gcc >=4.4 have loongson-specific support, we need to choose the suitable -march argument for gcc <= 4.3 and gcc >= 4.4, and here, we use -march=loongson2e for loongson2e. Thanks goes to Arnaud Patard <apatard@mandriva.com> for suggestion of using cc-options(Documentation/kbuild/makefiles.txt). and thanks Zhang Le for introducing the new CPU_LOONGSON2E kernel option. NOTE: -mtune option is not need if -march and -mtune use the same value. Signed-off-by: Zhang Le <r0bertz@gentoo.org> Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: Add a machtype kernel command line argumentWu Zhangjin2009-09-176-4/+53
| | | | | | | | | | | | | | | | | The difference between some loongson-based machines is very small, so, if there is no necessary to add new kernel config options to cope with this difference, it will be better to share the same kernel image file between them, benefit from this, the linux distribution developers only have a need to compile the kernel one time. This machtype kernel command line argument will be used later to share the same kernel image file between two different machines(menglong & yeeloong) made by lemote. Thanks very much to Zhang Le for cleaning up the machtype implementation. Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: Split common loongson source code outWu Zhangjin2009-09-1724-133/+351
| | | | | | | | | | | To share common loongson source code between all of the loongson-based machines. there is a need to split it out of the fuloong-2e/ directory. at the same time, other according tuning is needed. the machine-specific parts are defined as macros in relative header file, pci.h, mem.h, machine.h. Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: Change naming methodsWu Zhangjin2009-09-1725-90/+73
| | | | | | | | | | | | To make source code of loongson sharable to the machines(such as gdium) made by the other companies, we rename arch/mips/lemote to arch/mips/loongson, asm/mach-lemote to asm/mach-loongson, and rename lm2e to the name of the machine: fuloong-2e. accordingly, FULONG are renamed to FULOONG2E to make it distinguishable to the future FULOONG2F. and also, some other relative tuning is needed. Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: Add oprofile supportWu Zhangjin2009-09-175-0/+187
| | | | | | | | | | | | | | This kernel support is needed by the user-space tool:oprofile to profile linux kernel or applications via loongson2 performance counters. you can enable this driver via CONFIG_OPROFILE = y or m. On Loongson2 there are two performance counters, each one can count 16 events respectively. when anyone of the performance counter overflows, an interrupt will be generated and is routed to the IRQ MIPS_CPU_IRQ_BASE + 6. Signed-off-by: Yanhua <yanh@lemote.com> Signed-off-by: Wu Zhangjin <wuzj@lemote.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: RTC: Enable legacy RTC driver on fulongWu Zhangjin2009-09-171-1/+1
| | | | | | | | | | RTC_LIB is selected by MIPS by default, and therefore, the legacy RTC driver is disabled. but fortunately, RTC_LIB not works on fulong, so, enabling the legcy RTC driver is needed, otherwise, the tools like hwclock will not work. Signed-off-by: Wu Zhangjin <wuzj@lemote.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: PCI: Clean up pcimap setupWu Zhangjin2009-09-173-17/+45
| | | | | | | | Fixup the wrong original comment of pcimap, and make the source code more understandable. and also, some new extra consideration is added in. Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: clean up the coding styleWu Zhangjin2009-09-1714-194/+62
| | | | | | | | | | | | | | | | | | With the help of script/checkpatch.pl, i have cleaned up the coding style. 1. remove un-needed header files and tune some comments. 2. remove some un-needed { } add a new header file loongson.h: 3. move some common header files to loongson.h 4. move some common extern declartions to loongson.h and this new header file is needed for future loongson2f support. Signed-off-by: Wu Zhangjin <wuzj@lemote.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: Split the implementation of prom and setup partsWu Zhangjin2009-09-179-86/+168
| | | | | | | | | | | | | | | | | This patch split the old initilization and setup implementation to several file, one file one logic function. the other main changes include: 1. as the script/checkpatch.pl suggests, use strict_strtol instead of simple_strtol in arch/mips/lemote/lm2e/cmdline.c 2. use the existed macros in asm/mips-boards/bonito64.h as the arguments of set_io_port_base() and remove the un-needed ones in asm/mach-lemote/pci.h Signed-off-by: Wu Zhangjin <wuzj@lemote.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: PCI: use existing mips_io_port_baseWu Zhangjin2009-09-171-4/+1
| | | | | | | | mips_io_port_base is initialized via set_io_port_base() in arch/mips/lemote/lm2e/setup.c, we can use it directly here. Signed-off-by: Wu Zhangjin <wuzj@lemote.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: pm: clean up the reboot supportWu Zhangjin2009-09-171-8/+10
| | | | | | | | Several magic numbers have been replaced by relative macros, which will be more readable and understandable. Signed-off-by: Wu Zhangjin <wuzj@lemote.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: pm: Remove redundant source codeWu Zhangjin2009-09-171-6/+1
| | | | | | | | The implmentation of loongson2e_power_off and loongson2e_halt is almostly the same, just preserve one of them. Signed-off-by: Wu Zhangjin <wuzj@lemote.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: Add new early_printk implmentationWu Zhangjin2009-09-172-0/+46
| | | | | | | | | This patch is based on the implementation in the lm2e-fixes branch of Philippe's git://git.linux-cisco.org/linux-mips.git and the malta-specific early_printk implementation. Signed-off-by: Wu Zhangjin <wuzj@lemote.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: Remove out-of-date board-specific kgdb source codeWu Zhangjin2009-09-172-147/+1
| | | | | | | | | Since the re-implementation of kgdb by commit 8854700115ecf8aa6f087aa915b7b6cf18090d39 the platform-specific version has become superfluous, remove it. Signed-off-by: Wu Zhangjin <wuzj@lemote.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Loongson: Remove existing early_printk implementationWu Zhangjin2009-09-171-6/+0
| | | | | | | | | | | | Removes the existing implementation of early_printk for fulong. The old implementation was based on the outdated board-specific dbg_io.c. Since commit 8d60a903d986ffa26c41f0092320a3b9da20bfaf dbg_io.c is not longer needed; it will be removed by the next patch, and a new implementation of early_printk will be added later. Signed-off-by: Wu Zhangjin <wuzj@lemote.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: IP22, IP28: Build with -WerrorRalf Baechle2009-09-171-1/+1
| | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: FW: Build with -WerrorRalf Baechle2009-09-171-0/+2
| | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Fulong: Convert reset initialization to initcall.Ralf Baechle2009-09-173-6/+6
| | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Malta: Convert reset initialization to initcall.Ralf Baechle2009-09-172-2/+6
| | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Get rid of duplicate cpu_idle() prototype.Ralf Baechle2009-09-171-2/+1
| | | | | | Since 2.6.11-rc1 there is a prototype in <linux/smp.h>. Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: TXx9: Fix error handling.Julia Lawall2009-09-171-4/+13
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Error handling code following a kzalloc should free the allocated data. Error handling code following an ioremap should iounmap the allocated data. The semantic match that finds the first problem is as follows: (http://www.emn.fr/x-info/coccinelle/) // <smpl> @r exists@ local idexpression x; statement S; expression E; identifier f,f1,l; position p1,p2; expression *ptr != NULL; @@ x@p1 = \(kmalloc\|kzalloc\|kcalloc\)(...); ... if (x == NULL) S <... when != x when != if (...) { <+...x...+> } ( x->f1 = E | (x->f1 == NULL || ...) | f(...,x->f1,...) ) ...> ( return \(0\|<+...x...+>\|ptr\); | return@p2 ...; ) @script:python@ p1 << r.p1; p2 << r.p2; @@ print "* file: %s kmalloc %s return %s" % (p1[0].file,p1[0].line,p2[0].line) // </smpl> Signed-off-by: Julia Lawall <julia@diku.dk> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: TXx9: Disable PM capability of TX493[89] internal etherAtsushi Nemoto2009-09-171-0/+13
| | | | | | | | | Some TC35815 variants (i.e. TX493[89] internal ether) report existance of PM registers though they are not supported. Disable PM features by clearing pdev->pm_cap. Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: make page.h constants available to assembly.Nelson Elhage2009-09-171-2/+3
| | | | | | | | | | | page.h includes ifndef __ASSEMBLY__ guards, but PAGE_SIZE and some other constants are defined using "1UL", which the assembler does not support. Use the _AC macro from const.h to make them available to assembly (and linker scripts). Signed-off-by: Nelson Elhage <nelhage@ksplice.com> Signed-off-by: Tim Abbott <tabbott@ksplice.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Alchemy: add gpio_request/gpio_free stubs for CONFIG_GPIOLIB=nManuel Lauss2009-09-171-0/+9
| | | | | | | | | | Some drivers use gpio_request/gpio_free regardless of whether gpiolib is actually built; add stubs to work around the ensuing compile failures. Signed-off-by: Manuel Lauss <manuel.lauss@gmail.com> Tested-by: Florian Fainelli <florian@openwrt.org> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Lasat: Fix compilationAlexey Dobriyan2009-09-171-0/+1
| | | | | | | | | Header needed for current_cpu_data which expands to smp_processor_id(). However, linux/smp.h can't be included into asm/cpu-info.h due to horrible circular dependencies, so plug it here. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Octeon: Check all CCAs in cvmx_write_csr.David Daney2009-09-171-1/+1
| | | | | | | | | | | The current code only checks CCA of 0 when deciding if a dummy read is needed. Since the kernel can (and does) use other CCAs we need to mask out the CCA bits from the address. Since the address constant now fits in 16 bits, there is an added benefit that smaller code is generated. Signed-off-by: David Daney <ddaney@caviumnetworks.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Octeon: False positive timeoutRoel Kluin2009-09-171-1/+1
| | | | | | | | | | | | | | | If we reach the test just below the loop with a `timeout' value of 0, this does not mean that the timeout caused the loop to end, but rather the `smi_rd.s.pending', in the last iteration. If timeout caused the loop to end, then `timeout' is -1, not 0. Since this can occur only in the last iteration, it is not very likely to be a problem. By changing the post- to prefix decrement we ensure that a timeout of 0 does mean it timed out. Signed-off-by: Roel Kluin <roel.kluin@gmail.com> Acked-by: David Daney <ddaney@caviumnetworks.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* MIPS: Fix potencial build error in <asm/delay.h>Ralf Baechle2009-09-171-0/+2
| | | | | | | <asm/delay.h> will break if HZ isn't defined. In 2.6.26 and later we're usually lucky ... Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* Merge branch 'for-linus' of git://oss.sgi.com/xfs/xfsLinus Torvalds2009-09-1739-928/+593
|\ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * 'for-linus' of git://oss.sgi.com/xfs/xfs: (39 commits) xfs: includecheck fix for fs/xfs/xfs_iops.c xfs: switch to seq_file xfs: Record new maintainer information xfs: use correct log reservation when handling ENOSPC in xfs_create xfs: xfs_showargs() reports group *and* project quotas enabled xfs: un-static xfs_inobt_lookup xfs: actually enable the swapext compat handler xfs: simplify xfs_trans_iget xfs: merge fsync and O_SYNC handling xfs: speed up free inode search xfs: rationalize xfs_inobt_lookup* xfs: untangle xfs_dialloc xfs: factor out debug checks from xfs_dialloc and xfs_difree xfs: improve xfs_inobt_update prototype xfs: improve xfs_inobt_get_rec prototype xfs: factor out inode initialisation fs/xfs: Correct redundant test xfs: remove XFS_INO64_OFFSET un-static xfs_read_agf xfs: add more statics & drop some unused functions ...
| * Merge branch 'master' of git://oss.sgi.com/xfs/xfs into for-linusAlex Elder2009-09-1639-928/+593
| |\ | | | | | | | | | | | | Conflicts: fs/xfs/linux-2.6/xfs_lrw.c
| | * xfs: includecheck fix for fs/xfs/xfs_iops.cJaswinder Singh Rajput2009-09-151-1/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | fix the following 'make includecheck' warning: fs/xfs/linux-2.6/xfs_iops.c: xfs_acl.h is included more than once. Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com> Reviewed-by: Alex Elder <aelder@sgi.com> Signed-off-by: Alex Elder <aelder@sgi.com>
| | * xfs: switch to seq_fileAlexey Dobriyan2009-09-152-74/+55
| | | | | | | | | | | | | | | | | | | | | | | | create_proc_read_entry() is getting deprecated. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Reviewed-by: Alex Elder <aelder@sgi.com> Signed-off-by: Alex Elder <aelder@sgi.com>
| | * xfs: Record new maintainer informationAlex Elder2009-09-111-1/+1
| | | | | | | | | | | | | | | | | | | | | Update the MAINTAINERS file entry for XFS. Signed-off-by: Alex Elder <aelder@sgi.com> Signed-off by: Felix Blyakher <felixb@sgi.com>
| | * Merge branch 'master' of ↵Alex Elder2009-09-10438-3053/+5267
| | |\ | | | | | | | | | | | | git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6
| | * | xfs: use correct log reservation when handling ENOSPC in xfs_createChristoph Hellwig2009-09-101-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | We added the ENOSPC handling patch in xfs_create just after it got mered with xfs_mkdir. Change the log reservation to the variable for either the create or mkdir value so it does the right thing if get here for creating a directory. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
| | * | xfs: xfs_showargs() reports group *and* project quotas enabledAlex Elder2009-09-031-9/+13
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | If you enable group or project quotas on an XFS file system, then the mount table presented through /proc/self/mounts erroneously shows that both options are in effect for the file system. The root of the problem is some bad logic in the xfs_showargs() function, which is used to format the file system type-specific options in effect for a file system. The problem originated in this GIT commit: Move platform specific mount option parse out of core XFS code Date: 11/22/07 Author: Dave Chinner SHA1 ID: a67d7c5f5d25d0b13a4dfb182697135b014fa478 For XFS quotas, project and group quota management are mutually exclusive--only one can be in effect at a time. There are two parts to managing quotas: aggregating usage information; and enforcing limits. It is possible to have a quota in effect (aggregating usage) but not enforced. These features are recorded on an XFS mount point using these flags: XFS_PQUOTA_ACCT - Project quotas are aggregated XFS_GQUOTA_ACCT - Group quotas are aggregated XFS_OQUOTA_ENFD - Project/group quotas are enforced The code in error is in fs/xfs/linux-2.6/xfs_super.c: if (mp->m_qflags & (XFS_PQUOTA_ACCT|XFS_OQUOTA_ENFD)) seq_puts(m, "," MNTOPT_PRJQUOTA); else if (mp->m_qflags & XFS_PQUOTA_ACCT) seq_puts(m, "," MNTOPT_PQUOTANOENF); if (mp->m_qflags & (XFS_GQUOTA_ACCT|XFS_OQUOTA_ENFD)) seq_puts(m, "," MNTOPT_GRPQUOTA); else if (mp->m_qflags & XFS_GQUOTA_ACCT) seq_puts(m, "," MNTOPT_GQUOTANOENF); The problem is that XFS_OQUOTA_ENFD will be set in mp->m_qflags if either group or project quotas are enforced, and as a result both MNTOPT_PRJQUOTA and MNTOPT_GRPQUOTA will be shown as mount options. Signed-off-by: Alex Elder <aelder@sgi.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Felix Blyakher <felixb@sgi.com>
| | * | xfs: un-static xfs_inobt_lookupChristoph Hellwig2009-09-021-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | xfs_inobt_lookup is also used in xfs_itable.c, remove the STATIC modifier from it's declaration to fix non-debug builds. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Felix Blyakher <felixb@sgi.com> Signed-off-by: Felix Blyakher <felixb@sgi.com>
| | * | xfs: actually enable the swapext compat handlerChristoph Hellwig2009-09-011-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Fix a small typo in the compat ioctl handler that cause the swapext compat handler to never be called. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Torsten Kaiser <just.for.lkml@googlemail.com> Tested-by: Torsten Kaiser <just.for.lkml@googlemail.com> Reviewed-by: Eric Sandeen <sandeen@sandeen.net> Reviewed-by: Felix Blyakher <felixb@sgi.com> Signed-off-by: Felix Blyakher <felixb@sgi.com>
| | * | xfs: simplify xfs_trans_igetChristoph Hellwig2009-09-015-113/+5
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | xfs_trans_iget is a wrapper for xfs_iget that adds the inode to the transaction after it is read. Except when the inode already is in the inode cache, in which case it returns the existing locked inode with increment lock recursion counts. Now, no one in the tree every decrements these lock recursion counts, so any user of this gets a potential double unlock when both the original owner of the inode and the xfs_trans_iget caller unlock it. When looking back in a git bisect in the historic XFS tree there was only one place that decremented these counts, xfs_trans_iput. Introduced in commit ca25df7a840f426eb566d52667b6950b92bb84b5 by Adam Sweeney in 1993, and removed in commit 19f899a3ab155ff6a49c0c79b06f2f61059afaf3 by Steve Lord in 2003. And as long as it didn't slip through git bisects cracks never actually used in that time frame. A quick audit of the callers of xfs_trans_iget shows that no caller really relies on this behaviour fortunately - xfs_ialloc allows this inode from disk so it must not be there before, and all the RT allocator routines only every add each RT bitmap inode once. In addition to removing lots of code and reducing the size of the inode item this patch also avoids the double inode cache lookup in each create/mkdir/mknod transaction. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Alex Elder <aelder@sgi.com> Signed-off-by: Felix Blyakher <felixb@sgi.com>
| | * | xfs: merge fsync and O_SYNC handlingChristoph Hellwig2009-09-0110-112/+23
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The guarantees for O_SYNC are exactly the same as the ones we need to make for an fsync call (and given that Linux O_SYNC is O_DSYNC the equivalent is fdadatasync, but we treat both the same in XFS), except with a range data writeout. Jan Kara has started unifying these two path for filesystems using the generic helpers, and I've started to look at XFS. The actual transaction commited by xfs_fsync and xfs_write_sync_logforce has a different transaction number, but actually is exactly the same. We'll only use the fsync transaction going forward. One major difference is that xfs_write_sync_logforce never issues a cache flush unless we commit a transaction causing that as a side-effect, which is an obvious bug in the O_SYNC handling. Second all the locking and i_update_size vs i_update_core changes from 978b7237123d007b9fa983af6e0e2fa8f97f9934 never made it to xfs_write_sync_logforce, so we add them back. To make xfs_fsync easily usable from the O_SYNC path, the filemap_fdatawait call is moved up to xfs_file_fsync, so that we don't wait on the whole file after we already waited for our portion in xfs_write. We'll also use a plain call to filemap_write_and_wait_range instead of the previous sync_page_rang which did it in two steps including an half-hearted inode write out that doesn't help us. Once we're done with this also remove the now useless i_update_size tracking. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Felix Blyakher <felixb@sgi.com> Signed-off-by: Felix Blyakher <felixb@sgi.com>
| | * | xfs: speed up free inode searchDave Chinner2009-09-012-27/+115
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Don't search too far - abort if it is outside a certain radius and simply do a linear search for the first free inode. In AGs with a million inodes this can speed up allocation speed by 3-4x. [hch: ported to the new xfs_ialloc.c world order] Signed-off-by: Dave Chinner <dgc@sgi.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Alex Elder <aelder@sgi.com> Signed-off-by: Felix Blyakher <felixb@sgi.com>
| | * | xfs: rationalize xfs_inobt_lookup*Christoph Hellwig2009-09-013-71/+32
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Currenly we have a xfs_inobt_lookup* variant for each comparism direction, and all these get all three fields of the inobt records passed, while the common case is just looking for the inode number and we have only marginally more callers than xfs_inobt_lookup* variants. So opencode a direct call to xfs_btree_lookup for the single case where we need all fields, and replace xfs_inobt_lookup* with a xfs_inobt_looku that just takes the inode number and the direction for all other callers. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Alex Elder <aelder@sgi.com> Signed-off-by: Felix Blyakher <felixb@sgi.com>
| | * | xfs: untangle xfs_diallocChristoph Hellwig2009-09-011-153/+138
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Clarify the control flow in xfs_dialloc. Factor out a helper to go to the next node from the current one and improve the control flow by expanding composite if statements and using gotos. The xfs_ialloc_next_rec helper is borrowed from Dave Chinners dynamic allocation policy patches. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Alex Elder <aelder@sgi.com> Signed-off-by: Felix Blyakher <felixb@sgi.com>
| | * | xfs: factor out debug checks from xfs_dialloc and xfs_difreeDave Chinner2009-09-011-75/+56
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Factor out a common helper from repeated debug checks in xfs_dialloc and xfs_difree. [hch: split out from Dave's dynamic allocation policy patches] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Alex Elder <aelder@sgi.com> Signed-off-by: Felix Blyakher <felixb@sgi.com>
generated by cgit v1.2.3 (git 2.39.1) at 2025-01-09 18:22:35 +0100