Skip to contentSkip to conversation

2.6.6-rc1-mm1

DISCUSSION Andrew Morton 11 participants 30 messages Apr 19, 2004 – Apr 21, 2004

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.6-rc1/2.6.6-rc1-mm1/

- All of the anonmm rmap work is now merged up. No pte chains.

- Various cleanups and fixups, as usual.

- The list of external bk trees is getting a little short, due to problems at bkbits.net. The ones which are here are not necessarily very up-to-date with the various development trees.

Changes since 2.6.5-mm6:

linus.patch
bk-acpi.patch
bk-agpgart.patch
bk-alsa.patch
bk-cifs.patch
bk-cpufreq.patch
bk-drm.patch
bk-input.patch
bk-netdev.patch

External trees

-fix-mq_notify-with-sigev_none-notification.patch -radix-tree-comment-fix.patch -mq_open-and-close_on_exec.patch -rmap-4-flush_dcache-revisited.patch -rmap-5-swap_unplug-page.patch -rmap-6-nonlinear-truncation.patch -ext3-journalled-quotas.patch
-slab-panic.patch
-dm-fix-64-32-bit-ioctl-problems.patch -dm-check-the-uptodate-flag-in-sub-bios.patch -dm-handle-interrupts-within-suspend.patch -dm-use-wake_up-rather-than-wake_up_interruptible.patch -dm-log-an-error-if-the-target-has-unknown-target-type.patch -dm-correctly-align-the-dm_target_spec-structures.patch -dm-clarify-a-comment.patch -dm-retrieve_status-prevent-overrunning-the-ioctl-buffer.patch -dm-use-an-emit-macro.patch -kNFSdv4-4-of-10-nfsd4_readdir-fixes.patch -kNFSdv4-5-of-10-Fix-bad-error-returm-from-svcauth_gss_accept.patch -kNFSdv4-6-of-10-Keep-state-to-allow-replays-for-close-to-work.patch -kNFSdv4-7-of-10-Allow-locku-replays-aswell.patch -kNFSdv4-8-of-10-Improve-how-locking-copes-with-replays.patch -kNFSdv4-9-of-10-Set-credentials-properly-when-puutrootfh-is-used.patch -kNFSdv4-10-of-10-Implement-server-side-reboot-recovery-mostly.patch -kill-submit_bhbio-return-value.patch -pci-msi-kconfig-consolidation.patch -remove-buffer_error.patch -add-mqueue-support-to-x86-64.patch -light-weight-auditing-framework-for-s390.patch -posix-messages-queues-for-s390.patch -ppc64-fix-possible-duplicate-mmu-hash-entries.patch -fix-mq-32-bit-compatibility.patch -reiserfs-commit-default.patch -jbd-journal_dirty_metadata-locking-speedup.patch -sctp-printk-warnings.patch
-atm-warning-fixes.patch
-sir_dev-warnings.patch
-donauboe-ptr-fix.patch
-strip-warnings.patch
-strip-warnings-2.patch -print-warning-for-common-symbols-in-modules.patch -set_anon_super-locking-fix.patch

Merged

-kstrdup-and-friends.patch
-call_usermodehelper_async.patch
-call_usermodehelper_async-always.patch

Dropped, replaced with...

+create_singlethread_workqueue.patch

Extend workqueues so they don't unconditionally create one thread per cpu. (They still display as "khelper/0" in ps though..)

+use-workqueue-for-call_usermodehelper.patch

Use single-threaded workqueues for call_usermodehelper()

+ppc64-rtas-build-fix.patch

ppc64 compile fix

+rmap-7-object-based-rmap.patch
+ia64-rmap-build-fix.patch
+rmap-8-unmap-nonlinear.patch
+slab-panic.patch
+rmap-9-remove-pte_chains.patch
+rmap-10-add-anonmm-rmap.patch
+rmap-11-mremap-moves.patch
+rmap-12-pgtable-remove-rmap.patch
+rmap-13-include-asm-deletions.patch

The rest of anonmm-based rmap

-sched-config_sched_numa.patch

Dropped, wasn't popular.

+sched-fixes.patch

Scheduler tweaks

+nfs-direct-warning-fix.patch

Fix warning in nfs-O_DIRECT-fixes.patch

+numa-api-fixes.patch

Fixes to the NUMA API code

+numa-api-statistics-2.patch

Re-add statistical infrastructure to NUMA API.

These are currently undocumented. Hint.

+add-omitted-autofs4-super-block-field.patch

Fixup for the autofs4 patches

+autofs4-fix-handling-of-chdir-and-chroot.patch

Rework the autofs4 late mounting readdir support. Removes the racy handling in fs/open.c

+as-increase-batch-expiry.patch

Anticipatory scheduler tuning.

+direct-IO-retval-fix.patch
+direct-io-retval-fix-2.patch

Fix up the direct-IO code paths to correctly perform >2G I/O's on 64-bit architectures.

+populate-rootfs-later.patch

Call populate_rootfs() much later in boot, when the scheduler is actually ready for us to call schedule() without going splat.

+remove-amd7saucy_tco.patch

Remove defunct driver

+fix-default-value-for-commit-interval-for-older-reiserfs-filesystems.patch

Laptop-miode reiserfs fix

+consolidate-sys32_readv-and-sys32_writev.patch
+consolidate-do_execve32.patch
+consolidate-sys32_select.patch
+consolidate-sys32_nfsservctl.patch

Consolidate the code which performs emulation of readv & writev on 64-bit architectures.

+kill-off-hugepage_vma.patch

Remove hugepage_vma()

+ehci-oops-fix.patch

Fix some USB oops

+h8300-stack-bounds-checking.patch
+m68k-stack-bounds-checking.patch
+m68knommu-stack-bounds-checking.patch
+ppc32-stack-bounds-checking.patch
+sparc32-stack-bounds-checking.patch

Correctly calculate the top of stack in the backtracing code.

+3c509-oops-fix.patch

Fix an oops in 3c509.c

+lockfs-vfs-bits.patch
+lockfs-xfs-bits.patch
+lockfs-dm-bits.patch
+lockfs-dm-bits-2.patch

Move support for LVM snapshotting out of XFS and into core kernel.

+nfs-tokens-initdata.patch

Save a scrap of RAM.

+warn-if-module_param-and-module_parm-mixed.patch

Warn if a module uses both module_param() and the old MODULE_PARM()

+ide-cleanups-1.patch
+ide-cleanups-2.patch
+ide-cleanups-3.patch

IDE cleanups

+intel8x0-resume-fix.patch

Fix the intel8x0 driver for suspend/resume

+clean-up-asm-pgalloch-include.patch
+clean-up-asm-pgalloch-include-2.patch
+clean-up-asm-pgalloch-include-3.patch

Attempt to bring some sanity to the memory management include file ordering.

+ppc64-uninline-__pte_free_tlb.patch

Fix the ppc64 build for the above.

+lec-warning-fix.patch

Fix a warning in an ATM driver

All 212 patches:

linus.patch

create_singlethread_workqueue.patch
reate_singlethread_workqueue()

use-workqueue-for-call_usermodehelper.patch
Use workqueue for call_usermodehelper

ppc64-rtas-build-fix.patch
ppc64: rtas build fix

mc.patch
Add -mcN to EXTRAVERSION

bk-acpi.patch

bk-agpgart.patch

bk-alsa.patch

bk-cifs.patch

bk-cpufreq.patch

bk-drm.patch

bk-input.patch

bk-netdev.patch

mm.patch
add -mmN to EXTRAVERSION

kgdb-ga.patch
kgdb stub for ia32 (George Anzinger's one)
kgdbL warning fix
kgdb buffer overflow fix
kgdbL warning fix
kgdb: CONFIG_DEBUG_INFO fix
x86_64 fixes
correct kgdb.txt Documentation link (against 2.6.1-rc1-mm2)

kgdb-ga-recent-gcc-fix.patch
kgdb: fix for recent gcc

kgdboe-netpoll.patch
kgdb-over-ethernet via netpoll

kgdboe-configuration-logic-fix.patch
kgdboe: fix configuration of MAC address

kgdboe-configuration-logic-fix-fix.patch

kgdboe-non-ia32-build-fix.patch

kgdb-warning-fixes.patch
kgdb warning fixes

kgdb-x86_64-support.patch
kgdb-x86_64-support.patch for 2.6.2-rc1-mm3

kgdb-x86_64-warning-fixes.patch
kgdb-x86_64-warning-fixes

wchan-use-ELF-sections-kgdb-fix.patch
wchan-use-ELF-sections-kgdb-fix

kgdb-THREAD_SIZE-fixes.patch
THREAD_SIZE fixes for kgdb

rmap-7-object-based-rmap.patch
rmap 7 object-based rmap

ia64-rmap-build-fix.patch
ia64 rmap build fix

rmap-8-unmap-nonlinear.patch
rmap 8 unmap nonlinear

slab-panic.patch
slab: consolidate panic code

rmap-9-remove-pte_chains.patch
rmap 9 remove pte_chains

rmap-10-add-anonmm-rmap.patch
rmap 10 add anonmm rmap

rmap-11-mremap-moves.patch
rmap 11 mremap moves

rmap-12-pgtable-remove-rmap.patch
rmap 12 pgtable remove rmap

rmap-13-include-asm-deletions.patch
rmap 13 include/asm deletions

must-fix.patch
must fix lists update
must fix list update
mustfix update

must-fix-update-5.patch
must-fix update

ppc64-reloc_hide.patch

invalidate_inodes-speedup.patch
invalidate_inodes speedup
more invalidate_inodes speedup fixes

config_spinline.patch
uninline spinlocks for profiling accuracy.

pdflush-diag.patch

get_user_pages-handle-VM_IO.patch
fix get_user_pages() against mappings of /dev/mem

pci_set_power_state-might-sleep.patch

CONFIG_STANDALONE-default-to-n.patch
Make CONFIG_STANDALONE default to N

selinux-inode-race-trap.patch
Try to diagnose Bug 2153

slab-leak-detector.patch
slab leak detector
mm/slab.c warning in cache_alloc_debugcheck_after

local_bh_enable-warning-fix.patch

Move-saved_command_line-to-init-mainc.patch
Move saved_command_line to init/main.c

sched-run_list-cleanup.patch
small scheduler cleanup

sched-find_busiest_node-resolution-fix.patch
sched: improved resolution in find_busiest_node

sched-domains.patch
sched: scheduler domain support
sched: fix for NR_CPUS > BITS_PER_LONG
sched: clarify find_busiest_group
sched: find_busiest_group arithmetic fix

sched-find-busiest-fix.patch
sched-find-busiest-fix

sched-sibling-map-to-cpumask.patch
sched: cpu_sibling_map to cpu_mask
p4-clockmod sibling_map fix
p4-clockmod: handle more than two siblings

sched-domains-i386-ht.patch
sched: implement domains for i386 HT
sched: Fix CONFIG_SMT oops on UP
sched: fix SMT + NUMA bug
Change arch_init_sched_domains to use cpu_online_map Fix build with NR_CPUS > BITS_PER_LONG

sched-no-drop-balance.patch
sched: handle inter-CPU jiffies skew

sched-directed-migration.patch
sched_balance_exec(): don't fiddle with the cpus_allowed mask

sched-domain-debugging.patch
sched_domain debugging

sched-domain-balancing-improvements.patch
scheduler domain balancing improvements

sched-group-power.patch
sched-group-power
sched-group-power warning fixes

sched-domains-use-cpu_possible_map.patch
sched_domains: use cpu_possible_map

sched-smt-nice-handling.patch
sched: SMT niceness handling

sched-local-load.patch
sched: add local load metrics

process-migration-speedup.patch
Reduce TLB flushing during process migration

sched-trivial.patch
sched: trivial fixes, cleanups

sched-misc-fixes.patch
sched: misc fixes

sched-wakebalance-fixes.patch
sched: wakeup balancing fixes

sched-imbalance-fix.patch
sched: fix imbalance calculations

sched-altix-tune1.patch
sched: altix tuning

sched-fix-activelb.patch
sched: oops fix

ppc64-sched-domain-support.patch
ppc64: sched-domain support

sched-domain-setup-lock.patch
sched: fix setup races

ppc64-sched_domains-fix.patch
ppc64-sched_domains-fix

sched-domain-setup-lock-ppc64-fix.patch

sched-minor-cleanups.patch
sched: minor cleanups

sched-inline-removals.patch
sched: uninlinings

sched-move-cold-task.patch
sched: move cold task in mysteriouis ways

sched-migrate-shortcut.patch
sched: add migration shortcut

sched-more-sync-wakeups.patch
sched: extend sync wakeups

sched-boot-fix.patch
sched: lock cpu_attach_domain for hotplug

sched-cleanups.patch
sched: cleanups

sched-damp-passive-balance.patch
sched: passive balancing damping

sched-cpu-load-cleanup.patch
sched: cpu load management cleanup

sched-balance-context.patch
sched: balance-on-clone

sched-less-idle.patch
sched: reduce idle time

sched-wake_up-speedup.patch
sched: micro-optimisation for wake_up

sched-fixes.patch
sched: clean up leftovers

schedstats.patch
sched: scheduler statistics

fa311-mac-address-fix.patch
wrong mac address with netgear FA311 ethernet card

pid_max-fix.patch
Bug when setting pid_max > 32k

use-soft-float.patch
Use -msoft-float

non-readable-binaries.patch
Handle non-readable binfmt_misc executables

binfmt_misc-credentials.patch
binfmt_misc: improve calaulation of interpreter's credentials

aic7xxx-deadlock-fix.patch
aic7xxx deadlock fix

poll-select-longer-timeouts.patch
poll()/select(): support longer timeouts

poll-select-range-check-fix.patch
poll()/select() range checking fix

poll-select-handle-large-timeouts.patch
poll()/select(): handle long timeouts

add-a-slab-for-ethernet.patch
Add a kmalloc slab for ethernet packets

siimage-update.patch
ide: update for siimage driver

nmi_watchdog-local-apic-fix.patch
Fix nmi_watchdog=2 and P4 HT

nmi-1-hz-2.patch
reduce NMI watchdog call frequency with local APIC.

pcmcia-netdev-ordering-fixes.patch
PCMCIA netdevice ordering issues

3ware-update.patch
3ware driver update

idr-extra-features.patch
idr.c: extra features enhancements

shm-do_munmap-check.patch

stack-overflow-test-fix.patch
Fix stack overflow test for non-8k stacks

jbd-remove-livelock-avoidance.patch
JBD: remove livelock avoidance code in journal_dirty_data()

jgarzik-warnings.patch

logitech-keyboard-fix.patch
2.6.5-rc2 keyboard breakage

signal-race-fix.patch
signal handling race fix

signal-race-fix-ia64.patch
signal-race-fix: ia64

signal-race-fix-s390.patch
signal-race fixes for s390

signal-race-fix-x86_64.patch
signal-race-fixes: x86-64 support

signal-race-fixes-ppc.patch
signal-race fixes for ppc32 and ppc64

warn-on-mdelay-in-irq-handlers.patch
Warn on mdelay() in irq handlers

stack-reductions-nfsread.patch
stack reductions: nfs read

speed-up-sata.patch
speed up SATA

advansys-fix.patch
advansys check_region() fix

aic7xxx-unload-fix.patch
aic7xxx: fix oops whe hardware is not present

journal_add_journal_head-debug.patch
journal_add_journal_head-debug

nfs-O_DIRECT-fixes.patch
NFS: O_DIRECT fixes

nfs-direct-warning-fix.patch
nfs/direct.c warning fix

aic7xxx-swsusp-support.patch
support swsusp for aic7xxx

xfs-laptop-mode.patch
Laptop mode support for XFS

xfs-laptop-mode-syncd-synchronization.patch
Synchronize XFS sync daemon with laptop mode syncs.

list_del-debug.patch
list_del debug check

oops-dump-preceding-code.patch
i386 oops output: dump preceding code

lockmeter.patch
lockmeter
ia64 CONFIG_LOCKMETER fix

cciss-logical-device-queues.patch
cciss: per logical device queues

numa-api-x86_64.patch
numa api: -64 support

numa-api-bitmap-fix.patch
numa api: Bitmap bugfix

numa-api-i386.patch
numa api: Add i386 support

numa-api-ia64.patch
numa api: Add IA64 support

numa-api-core.patch
numa api: Core NUMA API code

mpol-in-copy_vma.patch
mpol in copy_vma

numa-api-core-slab-panic.patch
numa-api-core-slab-panic

numa-api-core-tweaks.patch
numa-api-core-tweaks

numa-api-fixes.patch
Some fixes for NUMA API

numa-api-statistics-2.patch
Re-add NUMA API statistics

numa-api-core-bitmap_clear-fixes.patch
numa-api-core bitmap_clear fixes

numa-api-vma-policy-hooks.patch
numa api: Add VMA hooks for policy

numa-api-vma-policy-hooks-fix.patch
numa-api-vma-policy-hooks fix

numa-api-shared-memory-support.patch
numa api: Add shared memory support

numa-api-shared-memory-support-tweaks.patch
numa-api-shared-memory-support-tweaks

numa-api-statistics.patch
numa api: Add statistics

numa-api-anon-memory-policy.patch
numa api: Add policy support to anonymous memory

sk98lin-buggy-vpd-workaround.patch
net/sk98lin: correct buggy VPD in ASUS MB
skvpd-build-fix

unplug-can-sleep.patch
unplug functions can sleep

fix-load_elf_binary-error-path-on-unshare_files-error.patch
fix load_elf_binary error path on unshare_files error

firestream-warnings.patch
firestream warnings

cpufreq_userspace-warning.patch
cpufreq_userspace warning

compute-creds-race-fix.patch
compute_creds race

compute-creds-race-fix-fix.patch
compute-creds-race-fix-fix

compute-creds-race-fix-fix-fix.patch
fix must_not_trace_exec() even more

rndis-fix.patch
usb/gadget/rndis.c fix

pc300_drv-warnings.patch
pc300_drv-warnings

fix-acer-travelmate-360-interrupt-routing.patch
fix Acer TravelMate 360 interrupt routing

ext3_rsv_cleanup.patch
ext3 block reservation patch set -- ext3 preallocation cleanup

ext3_rsv_base.patch
ext3 block reservation patch set -- ext3 block reservation

ext3_rsv_mount.patch
ext3 block reservation patch set -- mount and ioctl feature

ext3_rsv_dw.patch
ext3 block reservation patch set -- dynamically increase reservation window

ext3-reservation-default-on.patch
ext3 reservation: default to on

sched-in_sched_functions.patch
sched: in_sched_functions() cleanup

sysfs-d_fsdata-race-fix-2.patch
kobject/sysfs race fix

0-autofs4-2.6.0-signal-20040405.patch
autofs: dnotify + autofs may create signal/restart syscall loop

add-omitted-autofs4-super-block-field.patch
add omitted autofs4 super block field

1-autofs4-2.6.4-cleanup-20040405.patch
autofs: printk cleanups

2-autofs4-2.6.4-fill_super-20040405.patch

3-autofs4-2.6.0-bkl-20040405.patch
autofs: locking rework

4-autofs4-2.6.0-expire-20040405.patch
autofs: expiry refcount fixes

5-autofs4-2.6.0-readdir-20040405.patch
autofs: readdir fixes

umount-after-bad-chdir.patch
fix umount after bad chdir

autofs4-fix-handling-of-chdir-and-chroot.patch
autofs4: fix handling of chdir and chroot

6-autofs4-2.6.0-may_umount-20040405.patch
autofs: add ioctl to query unmountability

7-autofs4-2.6.0-extra-20040405.patch
autofs: readdir futureproofing

lindent-rwsem.patch
cleanup lib/rwsem.c lib/rwsem-spinlock.c

de-racify-rwsem-take-2.patch
de-racify rwsem take 2

scale-rwsem-take-2.patch
scale rwsem take 2

increase-number-of-dynamic-inodes-in-procfs-265.patch
Increase number of dynamic inodes in procfs

increase-number-of-dynamic-inodes-in-procfs-265-idr-init.patch
increase-number-of-dynamic-inodes-in-procfs-265-idr-init

ext3-data-leak-fix.patch
ext3 avoid writing kernel memory to disk

as-increase-batch-expiry.patch
AS: increase batch expiry intervals

direct-IO-retval-fix.patch
direct-IO-retval-fix

direct-io-retval-fix-2.patch
more direct-io retval fixes

populate-rootfs-later.patch
Call populate_rootfs later in boot

remove-amd7saucy_tco.patch
remove amd7xx_tco

fix-default-value-for-commit-interval-for-older-reiserfs-filesystems.patch
Fix default value for commit interval for older reiserfs filesystems.

consolidate-sys32_readv-and-sys32_writev.patch
Consolidate sys32_readv and sys32_writev

consolidate-do_execve32.patch
Consolidate do_execve32

consolidate-sys32_select.patch
Consolidate sys32_select

consolidate-sys32_nfsservctl.patch
Consolidate sys32_nfsservctl

kill-off-hugepage_vma.patch
From: David Gibson <[email protected]>
Subject: Kill off hugepage_vma()

ehci-oops-fix.patch
oops when loading ehci_hcd

h8300-stack-bounds-checking.patch
h8300 stack bounds checking

m68k-stack-bounds-checking.patch
m68k stack bounds checking

m68knommu-stack-bounds-checking.patch
m68knommu stack bounds checking

ppc32-stack-bounds-checking.patch
ppc32 stack bounds checking

sparc32-stack-bounds-checking.patch
sparc32 stack bounds checking

3c509-oops-fix.patch
3c509 oops fix

lockfs-vfs-bits.patch
lockfs - vfs bits

lockfs-xfs-bits.patch
lockfs - xfs bits

lockfs-dm-bits.patch
lockfs - dm bits

lockfs-dm-bits-2.patch
lockfs - dm bits

nfs-tokens-initdata.patch
nfs token table can be __initdata

warn-if-module_param-and-module_parm-mixed.patch
Warn if module_param and MODULE_PARM mixed

ide-cleanups-1.patch
IDE cleanups/fixups for 2.6.6-rc1 [1/3]

ide-cleanups-2.patch
IDE cleanups/fixups for 2.6.6-rc1 [2/3]

ide-cleanups-3.patch
IDE cleanups/fixups for 2.6.6-rc1 [3/3]

intel8x0-resume-fix.patch
intel8x0 suspend/resume fix

clean-up-asm-pgalloch-include.patch
Clean up asm/pgalloc.h include

clean-up-asm-pgalloch-include-2.patch
Clean up asm/pgalloc.h include

clean-up-asm-pgalloch-include-3.patch
Clean up asm/pgalloc.h include 3

ppc64-uninline-__pte_free_tlb.patch
ppc64: uninline __pte_free_tlb()

lec-warning-fix.patch
ATM warning fix

#1 lore raw
Quoted 6 lines · Andrew Morton

ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.6-rc1/2.6.6-rc1-mm1/
- All of the anonmm rmap work is now merged up. No pte chains.
- Various cleanups and fixups, as usual.
- The list of external bk trees is getting a little short, due to problems at bkbits.net. The ones which are here are not necessarily very up-to-date with the various development trees.

Okay, the cpumask_arith.h fixes aren't in here. What do I have to do to get the bare minimal correctness fixes in this area propagated to mainline?

The important aspect of these is that they're pertinent to small SMP systems, for instance, the dual Pee Cee shenanigans with all kinds of pins clipped along with all the other things used in larger boxen castrated.

-- wli

#2 lore raw
Quoted 9 lines · William Lee Irwin III
Quoted 6 lines · Andrew Morton

ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.6-rc1/2.6.6-rc1-mm1/
- All of the anonmm rmap work is now merged up. No pte chains.
- Various cleanups and fixups, as usual.
- The list of external bk trees is getting a little short, due to problems at bkbits.net. The ones which are here are not necessarily very up-to-date with the various development trees.

Quoted 7 lines

Okay, the cpumask_arith.h fixes aren't in here. What do I have to do to get the bare minimal correctness fixes in this area propagated to mainline?

The important aspect of these is that they're pertinent to small SMP systems, for instance, the dual Pee Cee shenanigans with all kinds of pins clipped along with all the other things used in larger boxen castrated.

I confess to being moderately exhasperated at the amount of talk and patching going on in the bitmap and cpumask areas. So when your patch floated past with a terse description which was bristling with ifs, buts and maybes I decided to take a pass.

If you want to send it again, cc'ing your co-conspirators and imparting some confidence that this darned thing is actually meandering toward a conclusion, please feel free ;)

#3 lore raw
Quoted 7 lines · Andrew Morton

I confess to being moderately exhasperated at the amount of talk and patching going on in the bitmap and cpumask areas. So when your patch floated past with a terse description which was bristling with ifs, buts and maybes I decided to take a pass.
If you want to send it again, cc'ing your co-conspirators and imparting some confidence that this darned thing is actually meandering toward a conclusion, please feel free ;)

Quick story is that what I sent is (what I believe) to be the bare minimum change to restore correctnes.

I'll start arguing with people to make sure bugfixes start moving and cleanups start waiting.

Paul, please remove akpm from the cc: list in future replies until we have come to a consensus and get this nailed down (hopefully ASAP) to a coherent cross-vendor story.

What I believe I have sent is the bare minimum change, with no cleanups or semantic changes. If you could review and/or send approval or the like that would be very helpful for the users of small SMP systems who are affected by the bug(s) you reported.

-- wli

#4 lore raw
Quoted 11 lines · William Lee Irwin III

Quick story is that what I sent is (what I believe) to be the bare minimum change to restore correctnes.
I'll start arguing with people to make sure bugfixes start moving and cleanups start waiting.
Paul, please remove akpm from the cc: list in future replies until we have come to a consensus and get this nailed down (hopefully ASAP) to a coherent cross-vendor story.
What I believe I have sent is the bare minimum change, with no cleanups or semantic changes. If you could review and/or send approval or the like that would be very helpful for the users of small SMP systems who are affected by the bug(s) you reported.

One last thing: I'm not opposed to cleanups in the least.

What I'm most concerned about is that arch maintainers have their needs satisfied by whatever internals you choose to back cpumask* with.

In all honesty, what you've been proposing is very close to what I wanted to have done to begin with. I would be glad to have you (or whoever else has the wherewithal to push the issue) get things down to the cleanest and most generally applicable implementation of API or definition of API as possible.

I have taken issue only where I believe you need to acquire arch maintainer feedback. IMHO, the changes proposed would be cleanups and more extensible, but only need the review from arch maintainers to bring them to full mainline merging. This is an API. When you have semantic equivalence, as approved by arch maintainers, this has no reason _not_ to go in.

All this said, you have pointed out immediate needs for fixes. Please let these fixes go through. There is a difference between the best code possible and what makes things work.

-- wli

#5 lore raw

bitmap, cpumask_arith (was: 2.6.6-rc1-mm1)

Ok ... lets see here.

This message is in reply to 4 messages from Bill Irwin, dated:
Sat, 17 Apr 2004 19:26:38 -0700
Sun, 18 Apr 2004 23:29:14 -0700
Sun, 18 Apr 2004 23:49:43 -0700
Mon, 19 Apr 2004 00:06:43 -0700

Quoted 1 line

I'm having enough trouble remembering what Paul Jackson's take on things

My take was that in my work-in-progress bitmap/cpumask patches, I have:

1) Slightly strengthened the constraints (post-conditions) on bitmap ops to not set tail bits so long as none were set on input. This mostly means that the complement operator zeros out the tail. The operators that return scalar (weight, for example) or boolean (empty and full, for example) are still careful to avoid depending on the state of the tail.

This changes Bill Irwin's "don't care" rule into a "don't care, but don't pollute further" rule.

2) Removed cpumask_arith entirely, along with 30 odd other files, to be replaced by a single cpumask.h set of macros, layered rather thinly on top of bitmap/bitop.

Quoted 2 lines

The important aspect of these is that they're pertinent to small SMP systems, ...

I tried making this point before, but failed to communicate. Let me try again. So far as I know, there is no instance in current Linux kernel code using this cpumask facility that is affected (currently broken) by the cpumask_arith bugs addressed by Bill's patch. Do you know of any breakage in other _current_ code caused by these cpumask_arith bugs, Bill?

As a consequence of my seeing nothing else yet overtly busted by this, it was, and remains, my recommendation to Bill to hold off on these patches. Little sense in correcting the semantics of a piece of code that I expected to remove shortly anyway, if nothing else cared for now.

However, I realize that Bill takes considerable pride in his work, and would like to see this fixed. It's not worth a big fuss, either way, so far as I can tell.

Quoted 1 line

Paul, please remove akpm from the cc: list

I have removed akpm from this reply.

Quoted 2 lines

the users of small SMP systems who are affected by the bug(s) you reported.

I am unaware that any user of any small SMP (or other) system is affected by these bugs.

Quoted 2 lines

If you could review and/or send approval or the like that would be very helpful ...

If we can agree on the actual state of affairs, then I will readily agree to such to Andrew, and let him decide on the merits and appropriate timing of this patch.

Quoted 1 line

From what I know now, this would mean telling Andrew, of Bill's patch:

This patch fixes some bugs in the small SMP system (2 to 32 CPU) flavor of cpumasks. From what we know now, these bugs aren't breaking any other existing code, but they are an accident waiting to happen. If Paul Jackson's bitmap/cpumask rework is adapted in the future, then this code being patched would be replaced anyway. But for now, these fixes are a definite improvement.

If this does not seem like a correct statement of affairs, lets hash that out. Then when we agree, you should present your patch to him on the lkml again, and I will gladly chime in with my endorsement.

Unless, of course, you change your mind and decide to hold off on this patch, to which I would even more readily agree <grin>.

Quoted 2 lines

I have taken issue only where I believe you need to acquire arch maintainer feedback.

And I have agreed that this is an excellent request. I was out on vacation last week, and hesitated to start the arch maintainer discussion the week before, knowing that I would be on vacation, and not wanting to start something I would not be around to follow through on.

But I am back now, and expect later this week to begin that discussion, as you have recommended.

Signature · 3 lines
                          I won't rest till it's the best ...
                          Programmer, Linux Scalability
                          Paul Jackson <[email protected]> 1.650.933.1373
#6 lore raw

Re: bitmap, cpumask_arith (was: 2.6.6-rc1-mm1)

[I sent this message 24 hours ago, but don't see it on lkml - try again. - pj]

Ok ... lets see here.

This message is in reply to 4 messages from Bill Irwin, dated:
Sat, 17 Apr 2004 19:26:38 -0700
Sun, 18 Apr 2004 23:29:14 -0700
Sun, 18 Apr 2004 23:49:43 -0700
Mon, 19 Apr 2004 00:06:43 -0700

Quoted 1 line

I'm having enough trouble remembering what Paul Jackson's take on things

My take was that in my work-in-progress bitmap/cpumask patches, I have:

1) Slightly strengthened the constraints (post-conditions) on bitmap ops to not set tail bits so long as none were set on input. This mostly means that the complement operator zeros out the tail. The operators that return scalar (weight, for example) or boolean (empty and full, for example) are still careful to avoid depending on the state of the tail.

This changes Bill Irwin's "don't care" rule into a "don't care, but don't pollute further" rule.

2) Removed cpumask_arith entirely, along with 30 odd other files, to be replaced by a single cpumask.h set of macros, layered rather thinly on top of bitmap/bitop.

Quoted 2 lines

The important aspect of these is that they're pertinent to small SMP systems, ...

I tried making this point before, but failed to communicate. Let me try again. So far as I know, there is no instance in current Linux kernel code using this cpumask facility that is affected (currently broken) by the cpumask_arith bugs addressed by Bill's patch. Do you know of any breakage in other _current_ code caused by these cpumask_arith bugs, Bill?

As a consequence of my seeing nothing else yet overtly busted by this, it was, and remains, my recommendation to Bill to hold off on these patches. Little sense in correcting the semantics of a piece of code that I expected to remove shortly anyway, if nothing else cared for now.

However, I realize that Bill takes considerable pride in his work, and would like to see this fixed. It's not worth a big fuss, either way, so far as I can tell.

Quoted 1 line

Paul, please remove akpm from the cc: list

I have removed akpm from this reply.

Quoted 2 lines

the users of small SMP systems who are affected by the bug(s) you reported.

I am unaware that any user of any small SMP (or other) system is affected by these bugs.

Quoted 2 lines

If you could review and/or send approval or the like that would be very helpful ...

If we can agree on the actual state of affairs, then I will readily agree to such to Andrew, and let him decide on the merits and appropriate timing of this patch.

Quoted 1 line

From what I know now, this would mean telling Andrew, of Bill's patch:

This patch fixes some bugs in the small SMP system (2 to 32 CPU) flavor of cpumasks. From what we know now, these bugs aren't breaking any other existing code, but they are an accident waiting to happen. If Paul Jackson's bitmap/cpumask rework is adapted in the future, then this code being patched would be replaced anyway. But for now, these fixes are a definite improvement.

If this does not seem like a correct statement of affairs, lets hash that out. Then when we agree, you should present your patch to him on the lkml again, and I will gladly chime in with my endorsement.

Unless, of course, you change your mind and decide to hold off on this patch, to which I would even more readily agree <grin>.

Quoted 2 lines

I have taken issue only where I believe you need to acquire arch maintainer feedback.

And I have agreed that this is an excellent request. I was out on vacation last week, and hesitated to start the arch maintainer discussion the week before, knowing that I would be on vacation, and not wanting to start something I would not be around to follow through on.

But I am back now, and expect later this week to begin that discussion, as you have recommended.

Signature · 3 lines
                          I won't rest till it's the best ...
                          Programmer, Linux Scalability
                          Paul Jackson <[email protected]> 1.650.933.1373
#7 lore raw

Re: bitmap, cpumask_arith (was: 2.6.6-rc1-mm1)

Quoted 5 lines · Paul Jackson

I tried making this point before, but failed to communicate. Let me try again. So far as I know, there is no instance in current Linux kernel code using this cpumask facility that is affected (currently broken) by the cpumask_arith bugs addressed by Bill's patch. Do you know of any breakage in other _current_ code caused by these cpumask_arith bugs,

Quoted 8 lines

Bill?
As a consequence of my seeing nothing else yet overtly busted by this, it was, and remains, my recommendation to Bill to hold off on these patches. Little sense in correcting the semantics of a piece of code that I expected to remove shortly anyway, if nothing else cared for now. However, I realize that Bill takes considerable pride in his work, and would like to see this fixed. It's not worth a big fuss, either way, so far as I can tell.

I thought you were reporting some instance affected by it. Since there's no report of anyone affected by it, and a cleanup/whatever pending, I'm dropping the bugfix patch(es).

-- wli

#8 lore raw
In reply to William Lee Irwin III
Quoted 10 lines · William Lee Irwin III
Quoted 6 lines · Andrew Morton

ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.6-rc1/2.6.6-rc1-mm1/
- All of the anonmm rmap work is now merged up. No pte chains.
- Various cleanups and fixups, as usual.
- The list of external bk trees is getting a little short, due to problems at bkbits.net. The ones which are here are not necessarily very up-to-date with the various development trees.

Quoted 3 lines

Okay, the cpumask_arith.h fixes aren't in here. What do I have to do to get the bare minimal correctness fixes in this area propagated to mainline?

Speaking of which, the CPU_MASK_ALL, CPU_MASK_NONE fix for cpumask_array.h still isn't there either.

It seems that in all the excitement a fix wasn't applied. Here is Linus' version, which is obviously the best one.

 linux-2.6-npiggin/include/asm-generic/cpumask_array.h |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff -puN include/asm-generic/cpumask.h~fix-big-cpumask include/asm-generic/cpumask.h

include/asm-generic/cpumask_array.h renamed from include/asm-generic/cpumask_array.h~fix-big-cpumask +2 -2
diff -puN include/asm-generic/cpumask_array.h~fix-big-cpumask include/asm-generic/cpumask_array.h
@@ -48,7 +48,7 @@
 /*
  * um, these need to be usable as static initializers
  */
-#define CPU_MASK_ALL	{ {[0 ... CPU_ARRAY_SIZE-1] = ~0UL} }
-#define CPU_MASK_NONE	{ {[0 ... CPU_ARRAY_SIZE-1] =  0UL} }
+#define CPU_MASK_ALL	((cpumask_t) { {[0 ... CPU_ARRAY_SIZE-1] = ~0UL} })
+#define CPU_MASK_NONE	((cpumask_t) { {[0 ... CPU_ARRAY_SIZE-1] =  0UL} })
 
 #endif /* __ASM_GENERIC_CPUMASK_ARRAY_H */

_

#9 lore raw

Re: 2.6.6-rc1-mm1 failure: kmod.o didn't compile with module-less setup

In reply to Andrew Morton

This one zonked out. I haven't configured module support, I simply compile a monolithic kernel tailored for this particular machine. I got this error, where 2.6.5-mm6 works well:

CC kernel/kmod.o
kernel/kmod.c: In function `call_usermodehelper':
kernel/kmod.c:253: error: `khelper_wq' undeclared (first use in this function) kernel/kmod.c:253: error: (Each undeclared identifier is reported only once kernel/kmod.c:253: error: for each function it appears in.) kernel/kmod.c: In function `usermodehelper_init':
kernel/kmod.c:267: error: `khelper_wq' undeclared (first use in this function) make[1]: *** [kernel/kmod.o] Error 1
make: *** [kernel] Error 2

Helge Hafting

#10 lore raw

Re: 2.6.6-rc1-mm1 failure: kmod.o didn't compile with module-less setup

Quoted 5 lines · Helge Hafting

This one zonked out. I haven't configured module support, I simply compile a monolithic kernel tailored for this particular machine. I got this error, where 2.6.5-mm6 works well:

Quoted 10 lines

CC kernel/kmod.o
kernel/kmod.c: In function `call_usermodehelper':
kernel/kmod.c:253: error: `khelper_wq' undeclared (first use in this function) kernel/kmod.c:253: error: (Each undeclared identifier is reported only once kernel/kmod.c:253: error: for each function it appears in.) kernel/kmod.c: In function `usermodehelper_init':
kernel/kmod.c:267: error: `khelper_wq' undeclared (first use in this function) make[1]: *** [kernel/kmod.o] Error 1
make: *** [kernel] Error 2

kernel/kmod.c renamed from kernel/kmod.c~use-workqueue-for-call_usermodehelper-fix +2 -1
diff -puN kernel/kmod.c~use-workqueue-for-call_usermodehelper-fix kernel/kmod.c
@@ -40,9 +40,10 @@
 
 extern int max_threads;
 
-#ifdef CONFIG_KMOD
 static struct workqueue_struct *khelper_wq;
 
+#ifdef CONFIG_KMOD
+
 /*
 	modprobe_path is set via /proc/sys.
 */

_

#11 lore raw

Re: 2.6.6-rc1-mm1 (compile stats)

In reply to Andrew Morton

Good run. Just have one new error in the allnoconfig compile (which developers tend not to run)...

CC kernel/kmod.o
kernel/kmod.c: In function `call_usermodehelper':
kernel/kmod.c:253: `khelper_wq' undeclared (first use in this function) kernel/kmod.c:253: (Each undeclared identifier is reported only once kernel/kmod.c:253: for each function it appears in.) kernel/kmod.c: In function `usermodehelper_init':
kernel/kmod.c:267: `khelper_wq' undeclared (first use in this function) make[1]: [kernel/kmod.o] Error 1 (ignored)

Linux 2.6 (mm tree) Compile Statistics (gcc 3.2.2) Warnings/Errors Summary

Kernel bzImage bzImage bzImage modules bzImage modules
(defconfig) (allno) (allyes) (allyes) (allmod) (allmod)
--------------- ---------- -------- -------- -------- -------- --------

2.6.6-rc1-mm1     0w/0e     0w/7e   122w/ 0e   7w/0e   4w/0e    122w/0e
2.6.5-mm6         0w/0e     0w/0e   123w/ 0e   7w/0e   4w/0e    124w/0e
2.6.5-mm5         0w/0e     0w/0e   119w/ 0e   7w/0e   4w/0e    120w/0e
2.6.5-mm4         0w/0e     0w/0e   120w/ 0e   7w/0e   4w/0e    121w/0e
2.6.5-mm3         0w/0e     1w/0e   121w/12e   7w/0e   3w/0e    123w/0e
2.6.5-mm2         0w/0e     0w/0e   128w/12e   7w/0e   3w/0e    134w/0e
2.6.5-mm1         0w/0e     5w/0e   122w/ 0e   7w/0e   3w/0e    124w/0e
2.6.5-rc3-mm4     0w/0e     0w/0e   124w/ 0e   8w/0e   4w/0e    126w/0e
2.6.5-rc3-mm3     0w/0e     5w/0e   129w/14e   8w/0e   4w/0e    129w/6e
2.6.5-rc3-mm2     0w/0e     5w/0e   130w/14e   8w/0e   4w/0e    129w/6e
2.6.5-rc3-mm1     0w/0e     5w/0e   129w/ 0e   8w/0e   4w/0e    129w/0e
2.6.5-rc2-mm5     0w/0e     5w/0e   130w/ 0e   8w/0e   4w/0e    129w/0e
2.6.5-rc2-mm4     0w/0e     5w/0e   134w/ 0e   8w/0e   3w/0e    133w/0e
2.6.5-rc2-mm3     0w/0e     5w/0e   134w/ 0e   8w/0e   3w/0e    133w/0e
2.6.5-rc2-mm2     0w/0e     5w/0e   137w/ 0e   8w/0e   3w/0e    134w/0e
2.6.5-rc2-mm1     0w/0e     5w/0e   136w/ 0e   8w/0e   3w/0e    134w/0e
2.6.5-rc1-mm2     0w/0e     5w/0e   135w/ 5e   8w/0e   3w/0e    133w/0e
2.6.5-rc1-mm1     0w/0e     5w/0e   135w/ 5e   8w/0e   3w/0e    133w/0e
2.6.4-mm2         1w/2e     5w/2e   144w/10e   8w/0e   3w/2e    144w/0e
2.6.4-mm1         1w/0e     5w/0e   146w/ 5e   8w/0e   3w/0e    144w/0e
2.6.4-rc2-mm1     1w/0e     5w/0e   146w/12e  11w/0e   3w/0e    147w/2e
2.6.4-rc1-mm2     1w/0e     5w/0e   144w/ 0e  11w/0e   3w/0e    145w/0e
2.6.4-rc1-mm1     1w/0e     5w/0e   147w/ 5e  11w/0e   3w/0e    147w/0e
2.6.3-mm4         1w/0e     5w/0e   146w/ 0e   7w/0e   3w/0e    142w/0e
2.6.3-mm3         1w/2e     5w/2e   146w/15e   7w/0e   3w/2e    144w/5e
2.6.3-mm2         1w/8e     5w/0e   140w/ 0e   7w/0e   3w/0e    138w/0e
2.6.3-mm1         1w/0e     5w/0e   143w/ 5e   7w/0e   3w/0e    141w/0e
2.6.3-rc3-mm1     1w/0e     0w/0e   144w/13e   7w/0e   3w/0e    142w/3e
2.6.3-rc2-mm1     1w/0e     0w/265e 144w/ 5e   7w/0e   3w/0e    145w/0e
2.6.3-rc1-mm1     1w/0e     0w/265e 141w/ 5e   7w/0e   3w/0e    143w/0e
2.6.2-mm1         2w/0e     0w/264e 147w/ 5e   7w/0e   3w/0e    173w/0e
2.6.2-rc3-mm1     2w/0e     0w/265e 146w/ 5e   7w/0e   3w/0e    172w/0e
2.6.2-rc2-mm2     0w/0e     0w/264e 145w/ 5e   7w/0e   3w/0e    171w/0e
2.6.2-rc2-mm1     0w/0e     0w/264e 146w/ 5e   7w/0e   3w/0e    172w/0e
2.6.2-rc1-mm3     0w/0e     0w/265e 144w/ 8e   7w/0e   3w/0e    169w/0e
2.6.2-rc1-mm2     0w/0e     0w/264e 144w/ 5e  10w/0e   3w/0e    171w/0e
2.6.2-rc1-mm1     0w/0e     0w/264e 144w/ 5e  10w/0e   3w/0e    171w/0e
2.6.1-mm5         2w/5e     0w/264e 153w/11e  10w/0e   3w/0e    180w/0e
2.6.1-mm4         0w/821e   0w/264e 154w/ 5e   8w/1e   5w/0e    179w/0e
2.6.1-mm3         0w/0e     0w/0e   151w/ 5e  10w/0e   3w/0e    177w/0e
2.6.1-mm2         0w/0e     0w/0e   143w/ 5e  12w/0e   3w/0e    171w/0e
2.6.1-mm1         0w/0e     0w/0e   146w/ 9e  12w/0e   6w/0e    171w/0e
2.6.1-rc2-mm1     0w/0e     0w/0e   149w/ 0e  12w/0e   6w/0e    171w/4e
2.6.1-rc1-mm2     0w/0e     0w/0e   157w/15e  12w/0e   3w/0e    185w/4e
2.6.1-rc1-mm1     0w/0e     0w/0e   156w/10e  12w/0e   3w/0e    184w/2e
2.6.0-mm2         0w/0e     0w/0e   161w/ 0e  12w/0e   3w/0e    189w/0e
2.6.0-mm1         0w/0e     0w/0e   173w/ 0e  12w/0e   3w/0e    212w/0e

Web page with links to complete details:
http://developer.osdl.org/cherry/compile/

John

#13 lore raw
Quoted 4 lines · Christoph Hellwig

4-autofs4-2.6.0-expire-20040405.patch exports vfsmount_lock which is probably not exactly a good design. It's only used by autofs4_may_umount which isn't autofs-specific at all.

Sorry Christoph, your recommendation is?

Ian

#14 lore raw
Quoted 10 lines · Ian Kent
Quoted 4 lines · Christoph Hellwig

4-autofs4-2.6.0-expire-20040405.patch exports vfsmount_lock which is probably not exactly a good design. It's only used by autofs4_may_umount which isn't autofs-specific at all.

Sorry Christoph, your recommendation is?

May as well rename that function to may_umount(), document it, suck it into fs/namespace.c or fs/namei.c and export it to modules.

That does increase the size of the static kernel a little, so arguably we shouldn't make this change until/unless we see a second user of the function.

#15 lore raw
Quoted 6 lines · Andrew Morton

May as well rename that function to may_umount(), document it, suck it into fs/namespace.c or fs/namei.c and export it to modules.

That does increase the size of the static kernel a little, so arguably we shouldn't make this change until/unless we see a second user of the function.

That's what I meant. Simply exporting vfsmount_lock to modules is a no-go.

#16 lore raw
Quoted 10 lines · Christoph Hellwig
Quoted 6 lines · Andrew Morton

May as well rename that function to may_umount(), document it, suck it into fs/namespace.c or fs/namei.c and export it to modules.

That does increase the size of the static kernel a little, so arguably we shouldn't make this change until/unless we see a second user of the function.

That's what I meant. Simply exporting vfsmount_lock to modules is a no-go.

Cool.

If it is decided to do this then would something like this be approiate Andrew?

I have
- renamed autofs4_may_umount to may_umount_tree and moved it to namespace.c
- removed the EXPORT_SYMBOL for vfsmount_lock
- updated autofs4 to suit

It is not possible to merge the functionality of may_umount into this as, it stands, as autofs v3 requires a slightly different semantic. That is if there are submounts that are not busy then it should return -EBUSY but may_umount_tree would return not busy.

fs/namespace.c +49 -2
@@ -37,8 +37,6 @@
 /* spinlock for vfsmount related operations, inplace of dcache_lock */
 spinlock_t vfsmount_lock __cacheline_aligned_in_smp = SPIN_LOCK_UNLOCKED;
 
-EXPORT_SYMBOL(vfsmount_lock);
-
 static struct list_head *mount_hashtable;
 static int hash_mask, hash_bits;
 static kmem_cache_t *mnt_cache; 
@@ -262,6 +260,55 @@
 	.show	= show_vfsmnt
 };
 
+/**
+ * may_umount_tree - check if a mount tree is busy
+ * @mnt: root of mount tree
+ *
+ * This is called to check if a tree of mounts has any
+ * open files, pwds or chroots. ie. if it is busy.
+ */
+int may_umount_tree(struct vfsmount *mnt)
+{
+	struct list_head *next;
+	struct vfsmount *this_parent = mnt;
+	int actual_refs;
+	int minimum_refs;
+
+	spin_lock(&vfsmount_lock);
+	actual_refs = atomic_read(&mnt->mnt_count);
+	minimum_refs = 2;
+repeat:
+	next = this_parent->mnt_mounts.next;
+resume:
+	while (next != &this_parent->mnt_mounts) {
+		struct vfsmount *p = list_entry(next, struct vfsmount, mnt_child);
+
+		next = next->next;
+
+		actual_refs += atomic_read(&p->mnt_count);
+		minimum_refs += 2;
+
+		if ( !list_empty(&p->mnt_mounts) ) {
+			this_parent = p;
+			goto repeat;
+		}
+	}
+
+	if (this_parent != mnt) {
+		next = this_parent->mnt_child.next;
+		this_parent = this_parent->mnt_parent;
+		goto resume;
+	}
+	spin_unlock(&vfsmount_lock);
+
+	if (actual_refs > minimum_refs)
+		return -EBUSY;
+
+	return 0;
+}
+
+EXPORT_SYMBOL(may_umount_tree);
+
 /*
  * Doesn't take quota and stuff into account. IOW, in some cases it will
  * give false negatives. The main reason why it's here is that we need
fs/autofs4/expire.c +1 -42

55 lines not on this page · raw message

#17 lore raw
Quoted 9 lines · On Wed, Apr 21, 2004 at 08:31:38PM +0800, [email protected] wrote:

If it is decided to do this then would something like this be approiate Andrew?

I have
- renamed autofs4_may_umount to may_umount_tree and moved it to namespace.c
- removed the EXPORT_SYMBOL for vfsmount_lock
- updated autofs4 to suit

Quoted 4 lines

It is not possible to merge the functionality of may_umount into this as, it stands, as autofs v3 requires a slightly different semantic. That is if there are submounts that are not busy then it should return -EBUSY but may_umount_tree would return not busy.

What about adding a paramter 'ignore_busy_submounts' and use most of the function in common for both autofs3 and autofs4?

Quoted 5 lines
fs/namespace.c +5 -0
+	struct list_head *next;
+	struct vfsmount *this_parent = mnt;
+	int actual_refs;
+	int minimum_refs;
+

fs/namespace.c line 286

Quoted 10 lines
fs/namespace.c +10 -0
+	spin_lock(&vfsmount_lock);
+	actual_refs = atomic_read(&mnt->mnt_count);
+	minimum_refs = 2;
+repeat:
+	next = this_parent->mnt_mounts.next;
+resume:
+	while (next != &this_parent->mnt_mounts) {
+		struct vfsmount *p = list_entry(next, struct vfsmount, mnt_child);
+
+		next = next->next;

Any chance to use list_for_each_entry here?

fs/namespace.c line 291

Quoted 1 line
fs/namespace.c +1 -0
+		if ( !list_empty(&p->mnt_mounts) ) {

This wants a white-space fixup.

#18 lore raw
Quoted 8 lines · Christoph Hellwig
Quoted 5 lines

It is not possible to merge the functionality of may_umount into this as, it stands, as autofs v3 requires a slightly different semantic. That is if there are submounts that are not busy then it should return -EBUSY but may_umount_tree would return not busy.

What about adding a paramter 'ignore_busy_submounts' and use most of the function in common for both autofs3 and autofs4?

Thought of that as I was writing the response.
I'll merge the two functions then and produce an updated patch.

Quoted 10 lines
Quoted 10 lines
fs/namespace.c +9 -0
+	struct list_head *next;
+	struct vfsmount *this_parent = mnt;
+	int actual_refs;
+	int minimum_refs;
+
+	spin_lock(&vfsmount_lock);
+	actual_refs = atomic_read(&mnt->mnt_count);
+	minimum_refs = 2;
+repeat:
Quoted 8 lines

fs/namespace.c line 286

Quoted 6 lines
fs/namespace.c +6 -0
+	next = this_parent->mnt_mounts.next;
+resume:
+	while (next != &this_parent->mnt_mounts) {
+		struct vfsmount *p = list_entry(next, struct vfsmount, mnt_child);
+
+		next = next->next;

Any chance to use list_for_each_entry here?

Same thought occured to me as well. As above.

Quoted 4 lines

fs/namespace.c line 291

Quoted 2 lines
fs/namespace.c +1 -0
+		if ( !list_empty(&p->mnt_mounts) ) {

This wants a white-space fixup.

Yes it's not quite the format I like most atm. I'll also do that.

May take a couple of days though.

Thanks for you continuing interest and help.

Ian

#19 lore raw
In reply to Christoph Hellwig
Quoted 5 lines · Christoph Hellwig
Quoted 5 lines
fs/namespace.c +5 -0
+
+	spin_lock(&vfsmount_lock);
+	actual_refs = atomic_read(&mnt->mnt_count);
+	minimum_refs = 2;
+repeat:
Quoted 8 lines

fs/namespace.c line 286

Quoted 6 lines
fs/namespace.c +6 -0
+	next = this_parent->mnt_mounts.next;
+resume:
+	while (next != &this_parent->mnt_mounts) {
+		struct vfsmount *p = list_entry(next, struct vfsmount, mnt_child);
+
+		next = next->next;

Any chance to use list_for_each_entry here?

It looks to me like this macro can't be used for a tree traversal. Please enlighten me if I'm wrong.

#20 lore raw
Quoted 4 lines · On Wed, Apr 21, 2004 at 11:52:18PM +0800, [email protected] wrote:
Quoted 1 line

Any chance to use list_for_each_entry here?

It looks to me like this macro can't be used for a tree traversal. Please enlighten me if I'm wrong.

Umm, indeed. The control flow in that function is a little too convoluted :)

#21 lore raw
In reply to Christoph Hellwig
Quoted 3 lines · Christoph Hellwig

That's what I meant. Simply exporting vfsmount_lock to modules is a no-go.

While I understand the motive for not exporting the lock the question of how one should obtain vfsmount structs when needed remains?

#22 lore raw
Quoted 2 lines · On Wed, Apr 21, 2004 at 08:39:59PM +0800, [email protected] wrote:

While I understand the motive for not exporting the lock the question of how one should obtain vfsmount structs when needed remains?

You shouldn't.

#23 lore raw
Quoted 6 lines · Christoph Hellwig
Quoted 2 lines · On Wed, Apr 21, 2004 at 08:39:59PM +0800, [email protected] wrote:

While I understand the motive for not exporting the lock the question of how one should obtain vfsmount structs when needed remains?

You shouldn't.

Shouldn't need them?

But your point is that they shouldn't need to be used and an different design is should be used, right.

Could make life hard for the automounter.
Possibly somewhat harder to solve the remaining limitations of autofs. But I haven't got a clear enough picture of what's needed yet (still).

I guess your point is that these services should reside in the VFS proper?

Ian

#24 lore raw
Quoted 8 lines · On Wed, Apr 21, 2004 at 09:52:01PM +0800, [email protected] wrote:
Quoted 6 lines · On Wed, Apr 21, 2004 at 08:39:59PM +0800, [email protected] wrote:
Quoted 2 lines

While I understand the motive for not exporting the lock the question of how one should obtain vfsmount structs when needed remains?

You shouldn't.

Shouldn't need them?

Exactly.

Quoted 8 lines

But your point is that they shouldn't need to be used and an different design is should be used, right.

Could make life hard for the automounter.
Possibly somewhat harder to solve the remaining limitations of autofs. But I haven't got a clear enough picture of what's needed yet (still).

I guess your point is that these services should reside in the VFS proper?

If you ask me much of what autofs does should reside in the VFS, namely triggering userspace upcalls as soon someone enters a special trigger (aka delayed mountpount) directory and expiry of vfsmounts.

#25 lore raw
Quoted 5 lines · Christoph Hellwig

If you ask me much of what autofs does should reside in the VFS, namely triggering userspace upcalls as soon someone enters a special trigger (aka delayed mountpount) directory and expiry of vfsmounts.

That's am approach that I've not had the luxury of pondering till now. I'll have to think about the potential of that for a while.

In the past I have thought that automount functionality is specialised enough to warrant seperation from the core VFS services. But now (after several months of consideration) I'm not sure that the functionality needed can be done without some general VFS support.

At the moment I think that if it was decided to add these services to the VFS then they would need to be general, not automount specific. As VFS services should be. But alas there are no clear requirements. For example, the recent proposal by Mike Waychison, although an excellent paper, requires a kernel expiry service but has no discussion of what is actually needed.

Sorry, I must be boring you with all this ranting.

Ian

#26 lore raw
In reply to Andrew Morton

Looking at the autofs4_may_umount I think I take the vfsmount_lock to late.

fs/autofs4/expire.c +1 -2
@@ -53,10 +53,9 @@
 	int actual_refs;
 	int minimum_refs;
 
+	spin_lock(&vfsmount_lock);
 	actual_refs = atomic_read(&mnt->mnt_count);
 	minimum_refs = 2;
-
-	spin_lock(&vfsmount_lock);
 repeat:
 	next = this_parent->mnt_mounts.next;
 resume:
#27 lore raw
In reply to Andrew Morton
Quoted 1 line · Andrew Morton

- All of the anonmm rmap work is now merged up. No pte chains.

Wonderful!

Signature · 3 lines
"Debugging is twice as hard as writing the code in the first place.
Therefore, if you write the code as cleverly as possible, you are,
by definition, not smart enough to debug it." - Brian W. Kernighan
#28 lore raw
Quoted 5 lines · Rik van Riel
Quoted 1 line · Andrew Morton

- All of the anonmm rmap work is now merged up. No pte chains.

Wonderful!

As is this (I assume debugging) message:

Apr 20 16:20:24 slagpiece kernel: xterm: mremap moved 49 cows

______
< Moo! >
------

        \   ^__^
         \  (oo)\_______
            (__)\       )\/\
                ||----w |
                ||     ||
#29 lore raw
Quoted 5 lines · Sean Neakums

As is this (I assume debugging) message:

Apr 20 16:20:24 slagpiece kernel: xterm: mremap moved 49 cows
Quoted 8 lines

______
< Moo! >
------

        \   ^__^
         \  (oo)\_______
            (__)\       )\/\
                ||----w |
                ||     ||

Lovely! Thanks for your report, and especially your art - you are the lucky winner of a huge churn of milk for first report of this message, just send your details etc. etc.

Yes, more or less a debugging message: nothing bad has happened, just something rather inefficient, and we'd like to get some idea of how much it occurs in practice. xterm, eh? I wouldn't have guessed that. Quite a herd.

I hoped a whimsical message might be reported more than a boring one!

Hugh