2.5.73-mjb1

2.5.73-mjb1

Post by Martin J. Blig » Thu, 26 Jun 2003 09:20:06



Right. If anyone breaks pfn_to_nid again, I'm going to reorganise
every damned include file under i386 - it's a sordid twisted mess ;-(

The patchset contains mainly scalability and NUMA stuff, and anything
else that stops things from irritating me. It's meant to be pretty stable,
not so much a testing ground for new stuff.

I'd be very interested in feedback from anyone willing to test on any
platform, however large or small.

ftp://ftp.kernel.org/pub/linux/kernel/people/mbligh/2.5.73/patch-2.5....

additional patches that can be applied if desired:

(these three form the qlogic feral driver)
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.5/2.5.72/...
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.5/2.5.72/...
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.5/2.5.72/...

Since 2.5.72-mjb2 (~ = changed, + = added, - = dropped)

Notes:

Now in Linus' tree:
- proc_lock                                     Martin J. Bligh / Andrew Morton
        Fix proc_lock handling and SDET hangs.
- numa_large_pages                              William Lee Irwin
        Place large pages in sensible places.
- sysfs_fix                                     Pat Mochel
        Fix something in sysfs, I forget what. Dave was complaining.

New:
+ uninitialised timer                           Mikael Pettersson
        Fix bogus cleanup so that gcc 2.95.4 works.
+ sysrq_t_fix                                   Andrew Morton
        Fix sysrq+t breakage where it showed the same stack for everyone
+ swsusp_state_check                            Matt Dobson
        Fix a check in s/w suspend code
+ pci_hotplug                                   Ivan K.
        Fix up pci hotplug w/o hotplug enabled.
+ pfn_to_nid                                    Martin J. Bligh
        Dance around the twisted rats nest of crap in i386 include.

Dropped:

- list-of-lists                                 Martin J. Bligh / Dave McCracken
        Causes more i_shared_sem contention. We're RCUing it ... later.

Pending:
Hyperthreaded scheduler (Ingo Molnar)
scheduler callers profiling (Anton or Bill Hartner)
Child runs first (akpm)
Kexec
e1000 fixes
Update the lost timer ticks code
pidmaps_nodepages (Dave Hansen)

Present in this patch:

early_printk                                    Dave Hansen / Keith Mannthey
        Allow printk before console_init

confighz                                        Andrew Morton / Dave Hansen
        Make HZ a config option of 100 Hz or 1000 Hz

config_page_offset                              Dave Hansen / Andrea
        Make PAGE_OFFSET a config option

numameminfo                                     Martin Bligh / Keith Mannthey
        Expose NUMA meminfo information under /proc/meminfo.numa

schedstat                                       Rick Lindsley
        Provide stats about the scheduler under /proc/schedstat

schedstat2                                      Rick Lindsley
        Provide more stats about the scheduler under /proc/schedstat

schedstat-scripts                               Rick Lindsley
        Provide some scripts for schedstat analysis under scripts/

sched_tunables                                  Robert Love
        Provide tunable parameters for the scheduler (+ NUMA scheduler)

irq_affinity                                    Martin J. Bligh
        Workaround for irq_affinity on clustered apic mode systems (eg x440)

partial_objrmap                                 Dave McCracken
        Object based rmap for filebacked pages.

kgdb                                            Andrew Morton
        The older version of kgdb, synched with 2.5.54-mm1

thread_info_cleanup (4K stacks pt 1)            Dave Hansen / Ben LaHaise
        Prep work to reduce kernel stacks to 4K

interrupt_stacks    (4K stacks pt 2)            Dave Hansen / Ben LaHaise
        Create a per-cpu interrupt stack.

stack_usage_check   (4K stacks pt 3)            Dave Hansen / Ben LaHaise
        Check for kernel stack overflows.

4k_stack            (4K stacks pt 4)            Dave Hansen
        Config option to reduce kernel stacks to 4K

4k_stacks_vs_kgdb                               Dave Hansen
        Fix interaction between kgdb and 4K stacks

stacks_from_slab                                William Lee Irwin
        Take kernel stacks from the slab cache, not page allocation.

thread_under_page                               William Lee Irwin
        Fix THREAD_SIZE < PAGE_SIZE case

spinlock_inlining                               Andrew Morton & Martin J. Bligh
        Inline spinlocks for profiling. Made into a ugly config option by me.

lockmeter                                       John Hawkes / Hanna Linder
        Locking stats.

reiserfs_dio                                    Mingming Cao
        DIO for Reiserfs

sched_interactive                               Ingo Molnar
        Bugfix for interactive scheduler

kgdb_cleanup                                    Martin J. Bligh
        Stop kgdb renaming schedule to do_schedule when it's not even enabled

acenic_fix                                      Martin J. Bligh
        Fix warning in acenic driver

local_balance_exec                              Martin J. Bligh
        Modify balance_exec to use node-local queues when idle

tcp_speedup                                     Martin J. Bligh
        Speedup TCP (avoid double copy) as suggested by Linus

disable preempt                                 Martin J. Bligh
        I broke preempt somehow, temporarily disable it to stop accidents

ppc64 fixes                                     Anton Blanchard
        Various PPC64 fixes / updates

numameminfo fix                                 Martin J. Bligh
        Correct /proc/meminfo.numa for zholes_size.

config_debug                                    Dave Hansen
        Make '-g' for the kernel a config option

akpm_bear_pit                                   Andrew Morton
        Add a printk for some buffer error I was hitting

32bit_dev_t                                     Andries Brouwer
        Make dev_t 32 bit

dynamic_hd_struct                               Badari Pulavarty
        Allocate hd_structs dynamically

lotsa_sds                                       Badari Pulavarty
        Create some insane number of sds

iosched_hashes                                  Badari Pulavarty
        Twiddle with the iosched hash tables for fun & profit

per_node_idt                                    Zwane Mwaikambo
        Per node IDT so we can do silly numbers of IO-APICs on NUMA-Q

config_numasched                                Dave Hansen
        Turn NUMA scheduler into a config option

lockmeter_tytso                                 Ted Tso
        Fix lockmeter

aiofix2                                         Mingming Cao
        fixed a bug in ioctx_alloc()

config_irqbal                                   Keith Mannthey
        Make irqbalance a config option

fs_aio_1_retry                                  Suparna Bhattacharya
        Filesystem aio. Chapter 1

fs_aio_2_read                                   Suparna Bhattacharya
        Filesystem aio. Chapter 2

fs_aio_3_write                                  Suparna Bhattacharya
        Filesystem aio. Chapter 3

fs_aio_4_down_wq                                Suparna Bhattacharya
        Filesystem aio. Chapter 4

fs_aio_5_wrdown_wq                              Suparna Bhattacharya
        Filesystem aio. Chapter 5

fs_aio_6_bread_wq                               Suparna Bhattacharya
        Filesystem aio. Chapter 6

fs_aio_7_ext2getblk_wq                          Suparna Bhattacharya
        Filesystem aio. Chapter 7

fs_aio_8_down_wq-ppc64                          Suparna Bhattacharya
        Filesystem aio. Chapter 8

fs_aio_9_down_wq-x86_64                         Suparna Bhattacharya
        Filesystem aio. Chapter 9

reslabify-pmd-pgd                               William Lee Irwin
        Stick things back in the slab. Or something.

separate_pmd                                    Dave Hansen
        Separate kernel pmd per task.

banana_split                                    Dave Hansen
        Make PAGE_OFFSET play twister and limbo.

percpu_real_loadavg                             Dave Hansen / Martin J. Bligh
        Tell me what the real load average is, and tell me per cpu.

nolock                                          Dave McCracken
        Nah, we don't like locks.

proc_pid_readdir                                Manfred Spraul
        Make proc_pid_readdir more efficent. Allegedly.

mbind_part1                                     Matt Dobson
        Bind some memory for NUMA.

mbind_part2                                     Matt Dobson
        Bind some more memory for NUMA.

per_node_rss                                    Matt Dobson
        Track which nodes tasks mem is on, so sched can be sensible.

uninitialised timer                             Mikael Pettersson
        Fix bogus cleanup so that gcc 2.95.4 works.

sysrq_t_fix                                     Andrew Morton
        Fix sysrq+t breakage where it showed the same stack for everyone

swsusp_state_check                              Matt Dobson
        Fix a check in s/w suspend code

pci_hotplug                                     Ivan K.
        Fix up pci hotplug w/o hotplug enabled.

pfn_to_nid                                      Martin J. Bligh
        Dance around the twisted rats nest of crap in i386 include.

-mjb                                            Martin J. Bligh
        Add a tag to the makefile

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/