BitKeeper patches since v2.6.0: 145 Changesets

Gzipped full patch from v2.6.0 to ChangeSet 1.1506

ChangeSet 2003/12/19 16:20:59-08:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/12/19 16:18:47-08:00, tony.luck @ intel.com [diffview]

[PATCH] ia64: enable recovery from TLB errors

Here's the updated version of the MCA TLB recovery patch.

ChangeSet 2003/12/19 16:11:57-08:00, kaos @ sgi.com [diffview]

[PATCH] ia64: fix deadlock in ia64_mca_cmc_int_caller()

smp_call_function() must not be called from interrupt context (can
deadlock on tasklist_lock).  Use keventd to call smp_call_function().

ChangeSet 2003/12/19 16:10:05-08:00, kaos @ sgi.com [diffview]

[PATCH] ia64: sync pal/sal/salinfo/mca with 2.4 code

Forward port the recent changes to pal.h, sal.h, mca.h, salinfo.c and
mca.c from 2.4.23-rc2 to 2.6.0-test9.

This converts 2.6 to use salinfo instead of printing CMC/CPE/MCA/INIT
records in the kernel.  It makes the two kernel versions as close
together as possible.

ChangeSet 2003/12/19 15:59:23-08:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/12/19 15:40:20-08:00, jbarnes @ sgi.com [diffview]

[PATCH] ia64: sn2 defconfig file

As promised, here's a patch to add an sn2 defconfig file to get people
started with 2.6 kernels.  I even turned on CONFIG_IA64_SGI_SIM support
to make Jack happy :)

ChangeSet 2003/12/19 15:35:51-08:00, jbarnes @ sgi.com [diffview]

[PATCH] ia64: initialize bootmem maps in reverse order

The arch-independent bootmem code now requires that arches initialize
their bootmem maps in reverse order (in particular, from high to low
addesses), otherwise alloc_bootmem_pages_low() won't work.  This change
makes the ia64 code do just that, so that machines without an IOMMU can
allocate their bounce buffers in low memory at early boot.  It also adds
a sanity check to the early init code to make sure that each node has a
local data area, because if they don't, many things will break later on
and may be hard to track down.

ChangeSet 2003/12/19 15:20:29-08:00, bjorn.helgaas @ hp.com [diffview]

[PATCH] ia64: Prevent SAL calls from being preempted

(SAL_CALL_REENTRANT): Disable preemption around the SAL call to
make sure we don't get rescheduled on a different CPU.

ChangeSet 2003/12/19 15:19:55-08:00, bjorn.helgaas @ hp.com [diffview]

[PATCH] ia64: Force generic and hp kernels to use 16MB granules

This forces the granule size to 16MB for HP zx1 and generic
kernels.  HP sx1000 machines require this.

ChangeSet 2003/12/19 15:18:47-08:00, bjorn.helgaas @ hp.com [diffview]

[PATCH] ia64: Remove unused ACPI functions.

Remove unused functions:
    acpi_get_prt()
    acpi_get_interrupt_model()
    acpi_get_addr_space()

ChangeSet 2003/12/19 15:17:48-08:00, bjorn.helgaas @ hp.com [diffview]

[PATCH] ia64: Remove extraneous printks (we get the same information from ACPI).

(iosapic_init): Remove extraneous printk.
(pci_acpi_scan_root): Remove extraneous printk.

ChangeSet 2003/12/19 15:16:43-08:00, bjorn.helgaas @ hp.com [diffview]

[PATCH] ia64: Fix PCI root bridge resources to handle prior allocations.

(alloc_resources):  Use insert_resource(), not request_resource(), to
allocate PCI root bridge windows.  This fixes the problem
where root bridge window allocation fails because an early
driver (like VGA) has already allocated things.

ChangeSet 2003/12/19 15:15:20-08:00, jbarnes @ sgi.com [diffview]

[PATCH] ia64: make NODES_SHIFT a little biggger

Make NODES_SHIFT larger to accomodate 256 node machines.

ChangeSet 2003/12/19 15:14:29-08:00, jbarnes @ sgi.com [diffview]

[PATCH] ia64: update sn2 MAINTAINERS file entry

ChangeSet 2003/12/19 15:13:48-08:00, jbarnes @ sgi.com [diffview]

[PATCH] ia64: make cpu_to_node_map unsigned

This small fix is needed for machines with more than 128 nodes.

ChangeSet 2003/12/19 15:09:49-08:00, steiner @ sgi.com [diffview]

[PATCH] ia64: prevent buffer-overrun in acpi_numa_memory_affinity_init()

The code in acpi_numa_memory_affinity_init that sorts the node_memblk
can overrun the array & clobber the memory that follows the end of the
array. The error will be seen only on systems that fill the
node_memblk array and only if SAL doesnt sort the entries in the SRAT.

ChangeSet 2003/12/19 11:48:09-08:00, peterc @ gelato.unsw.edu.au [diffview]

[PATCH] ia64: enable out-of-tree compilation for IA64

ChangeSet 2003/12/17 22:53:18-08:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/12/16 09:17:33-08:00, davidm @ tiger.hpl.hp.com [diffview]

ia64: Bring export of spin-lock contention-routines in sync with
        this change:  Jim Wilson says that gcc v3.3 also supports
        marking ar.pfs as clobbered, so use ia64_spinlock_contention()
        for any GCC with v3.3 or newer.

ChangeSet 2003/12/13 01:04:41-08:00, davidm @ tiger.hpl.hp.com [diffview]

ia64: Fix compiler warning in intrinsics.h.

ChangeSet 2003/12/13 00:56:49-08:00, davidm @ tiger.hpl.hp.com [diffview]

ia64: Fix ivt overflow that occurred when turning on
        CONFIG_DISABLE_VHPT.

ChangeSet 2003/12/13 00:04:17-08:00, davidm @ tiger.hpl.hp.com [diffview]

ia64: Fix bug discovered by Bill Nottingham & Jakub Jelinek where
        put_user() arguments with function-calls would cause the
        macro to return unexpected values.  Fixed by avoiding macro
        argument evaluation while r8/r9 are in use for exception-handling
        purposes.  Also, consolidated access-macros so that GCC and
        Intel compiler use 90% the same code.

ChangeSet 2003/12/11 12:23:26-08:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/12/11 12:22:16-08:00, davidm @ tiger.hpl.hp.com [diffview]

ia64: Based on patch by Jerome Marchand: Add ia64-optimized
        atomic_dec_and_lock().  Actually, this could be the generic
        version for any platform that has cmpxchg().

ChangeSet 2003/12/10 17:34:27-08:00, davidm @ tiger.hpl.hp.com [diffview]

ia64: Switch places for the gate pages and the guard page.  This improves
        backwards-compatibility with older (broken) versions of GCC which
        recognize a signal-handler only if it is in the address range
        from 0xa000000000000100. to 0xa000000000020000.

ChangeSet 2003/12/10 17:29:05-08:00, davidm @ tiger.hpl.hp.com [diffview]

ia64: Jim Wilson says that gcc v3.3 also supports marking ar.pfs as
        clobbered, so use ia64_spinlock_contention() for any GCC with
        v3.3 or newer.

ChangeSet 2003/12/08 17:37:01-08:00, peter @ chubb.wattle.id.au [diffview]

[PATCH] ia64: make perfmon CONFIG_PREEMPT-safe again

Here's a  fix for non-preemption safety in perfmon.c.

I haven't tried it while running a preemption stress test, but this
allows q-syscollect to work.

ChangeSet 2003/12/04 14:53:04-08:00, jakub @ redhat.com [diffview]

[PATCH] ia64: fix typo in vmlinux.lds.S

The security init section was incorrectly using PAGE_OFFSET instead of
LOAD_OFFSET.

ChangeSet 2003/12/01 13:13:14-08:00, pj @ sgi.com [diffview]

[PATCH] ia64: fix smp_affinity user-space accesses

Here is a new improved patch for verifying user access to string
passed in to kernel on write to /proc/irq/<pid>/smp_affinity.

The access_ok() but missing __get_user() on each byte earlier patch
has been replaced with a copy_from_user().

I have built it and verified that it handles write requests
as before, on an ia64 system (well - you can no longer pass
more than 14 spaces after the 'R' - tough).

ChangeSet 2003/11/26 20:36:42-08:00, davidm @ tiger.hpl.hp.com [diffview]

Drop KERN_CACHEDECAYTICKS sysctl.
Add KERN_GATE_DSO sysctl, which allows to register a pathname for
the file containing an image of the gate DSO.

ChangeSet 2003/11/26 15:03:24-08:00, davidm @ tiger.hpl.hp.com [diffview]

ia64: Fix a bug in sigtramp() which corrupted ar.rnat when unwinding
        across a signal trampoline (in user space).  Reported by
        Laurent Morichetti.

ChangeSet 2003/11/26 14:19:26-08:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/11/24 10:15:25-08:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/10/27 11:53:59-08:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/10/24 17:15:49-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/10/21 12:37:15-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/10/20 23:05:00-07:00, davidm @ tiger.hpl.hp.com [diffview]

ia64: Make drivers/acpi/tables.c compile again.

ChangeSet 2003/10/20 16:26:48-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/10/20 14:20:09-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/10/17 23:00:05-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/10/16 11:52:43-07:00, bjorn.helgaas @ hp.com [diffview]

[PATCH] early_serial_setup array bounds check

early_serial_setup() doesn't validate the array index,
so a caller could corrupt memory after serial8250_ports[]
by supplying a value of port->line that's too large.

I haven't seen a failure related to this, but it seems fragile
to rely on callers to know how many ports the driver supports.

ChangeSet 2003/10/16 11:51:12-07:00, bjorn.helgaas @ hp.com [diffview]

[PATCH] serial console registration bugfix

uart_set_options() can dereference a null pointer.  This happens
if you specify a console that hasn't previously been setup by
early_serial_setup().

For example, on ia64, the HCDP typically tells us about line 0,
so we calls early_serial_setup() for it.  If the user specifies
"console=ttyS3", we machine-check when trying to follow the
uninitialized port->ops pointer.

It's not entirely clear to me whether we should return 0 or -ENODEV
or something.  The advantage of returning zero is that if the user
specifies "console=ttyS0" and we just lack the HCDP, the console
doesn't work as early as usual, but it does start working after the
serial driver detects the port (though the baud/parity/etc from the
command line are lost).  Returning -ENODEV seems to prevent it from
ever working.

ChangeSet 2003/10/15 10:18:45-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/10/10 16:31:19-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/10/09 16:33:26-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge with 2.6.0-test7.

ChangeSet 2003/09/23 13:59:07-07:00, davidm @ tiger.hpl.hp.com [diffview]

Make NFS server work on ia64 (and other 64-bit arches) again.
Patch received via Matthew Wilcox.

ChangeSet 2003/09/23 11:44:31-07:00, davidm @ wailua.hpl.hp.com [diffview]

Remove bogus __attribute_used__ in module.h.

ChangeSet 2003/09/18 12:17:08-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/09/18 12:13:43-07:00, jbarnes @ sgi.com [diffview]

[PATCH] ia64: fix early printk for sn2

Quick fix for the early printk function name for sn2.  Goes along with a
patch I'm about to send to Andrew.

ChangeSet 2003/09/08 23:24:59-07:00, davidm @ tiger.hpl.hp.com [diffview]

Some more __attribute_used__ fixes.

ChangeSet 2003/09/08 16:19:54-07:00, davidm @ tiger.hpl.hp.com [diffview]

Initial merge with 2.6.0-test5.

ChangeSet 2003/09/08 13:26:34-07:00, davidm @ tiger.hpl.hp.com [diffview]

Mark initcall macros with __attribute_used__ so the definitions do not
get optimized away by the compiler (such as the latest GCC pre-3.4).

ChangeSet 2003/09/05 18:40:04-07:00, bjorn.helgaas @ hp.com [diffview]

[PATCH] ia64: Add HP AGP 8X bridge and fix ACPI claim

The following patch to the HP ZX1 GART driver
- adds recognition of the HWP0007 device (an AGP 8X bridge)
- fixes the use of acpi_get_devices()

The problem with acpi_get_devices() is that it always returns AE_OK,
except when the callback (zx1_gart_probe()) returns a failure.  That
means agp_hp_init() cannot use the status from acpi_get_devices()
to distinguish between (1) an AGP bridge was found and successfully
initialized, and (2) no AGP bridge was found at all.

ChangeSet 2003/08/28 22:19:15-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/08/28 16:20:08-07:00, rusty @ rustcorp.com.au [diffview]

[PATCH] Fix Slowdown in /proc/kallsyms

Out by one error broke caching of results in /proc/kallsyms, slowing
reading to a crawl.

ChangeSet 2003/08/25 15:39:44-07:00, davidm @ tiger.hpl.hp.com [diffview]

Patch by Arun to make test-popseg.x86 work again: I've tracked this down to
mm/memory.c:do_no_page() dereferencing vma->vm_file. The problem is that
the vma used for mapping the GDT has vma->vm_ops->nopage, but vma->vm_file
is NULL.

Andrew Morton agreed with this patch, but for some reason, it didn't make
it into 2.6.0-test4.

ChangeSet 2003/08/25 12:39:40-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/08/25 10:09:16-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/08/21 16:00:35-07:00, davidm @ tiger.hpl.hp.com [diffview]

Manual merge.

ChangeSet 2003/08/19 23:25:23-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/08/19 16:58:14-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/08/19 15:56:00-07:00, davidm @ wailua.hpl.hp.com [diffview]

(kallsyms_open): Fix oops caused by uninitialized memory.

ChangeSet 2003/08/15 18:34:07-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/08/15 13:46:29-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge with 2.6.0-test3.

ChangeSet 2003/08/07 16:18:34-07:00, davidm @ tiger.hpl.hp.com [diffview]

tables.c:
  Check for X__DSDT in acpi_get_table_header_early()

ChangeSet 2003/08/07 14:39:48-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/07/29 00:26:02-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/07/29 00:02:16-07:00, davidm @ tiger.hpl.hp.com [diffview]

ia64: Merge with 2.6.0-test2.

ChangeSet 2003/07/25 18:38:59-07:00, davidm @ tiger.hpl.hp.com [diffview]

Update with latest version from Linus' tree.

ChangeSet 2003/07/15 14:47:35-07:00, rusty @ rustcorp.com.au [diffview]

[PATCH] Use local_t for module reference counts

ChangeSet 2003/07/15 11:25:20-07:00, davidm @ tiger.hpl.hp.com [diffview]

Cset exclude: davidm@tiger.hpl.hp.com|ChangeSet|20030711020407|51156
Cset exclude: davidm@tiger.hpl.hp.com|ChangeSet|20030709222337|34756

ChangeSet 2003/07/14 21:58:38-07:00, davidm @ tiger.hpl.hp.com [diffview]

Undo (most of) the ia64-specific changes to pci_irq.c.
iosapic_parse_prt() is the right place to do those.

ChangeSet 2003/07/14 17:15:53-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge with 2.6.0-pre1.

ChangeSet 2003/07/12 01:56:01-07:00, davidm @ tiger.hpl.hp.com [diffview]

Remove extra blank line.

ChangeSet 2003/07/12 01:06:08-07:00, davidm @ tiger.hpl.hp.com [diffview]

Clean up some no-longer-needed ia64 workarounds.

ChangeSet 2003/07/10 18:38:35-07:00, davidm @ tiger.hpl.hp.com [diffview]

Move EARLY_PRINTK config options to this repository, where the code lives.

ChangeSet 2003/07/10 18:30:20-07:00, davidm @ tiger.hpl.hp.com [diffview]

ia64: Merge with 2.5.75.

ChangeSet 2003/07/09 11:31:00-07:00, davidm @ tiger.hpl.hp.com [diffview]

Manual merge with 2.5.74+.

ChangeSet 2003/06/26 15:43:59-07:00, davidm @ tiger.hpl.hp.com [diffview]

Cset exclude: davidm@tiger.hpl.hp.com|ChangeSet|20030621161908|14888
Cset exclude: davidm@tiger.hpl.hp.com|ChangeSet|20030621161905|15263

ChangeSet 2003/06/23 23:34:01-07:00, davidm @ wailua.hpl.hp.com [diffview]

osl.c:
  (acpi_os_readable): Correct type of 2nd arg to acpi_size.

ChangeSet 2003/06/23 23:09:51-07:00, davidm @ tiger.hpl.hp.com [diffview]

Sync with Linus' tree.

ChangeSet 2003/06/23 17:16:54-07:00, davidm @ tiger.hpl.hp.com [diffview]

Update ia64 defconfig.

ChangeSet 2003/06/19 13:50:01-07:00, davidm @ tiger.hpl.hp.com [diffview]

More ppc64 bad merge fixing...

ChangeSet 2003/06/19 13:44:03-07:00, tony.luck @ intel.com [diffview]

[PATCH] ia64: add missing __exit_p() to i460-agp.c

ChangeSet 2003/06/19 13:17:49-07:00, davidm @ tiger.hpl.hp.com [diffview]

Drop export of init_thread_union alltogether.

ChangeSet 2003/06/19 13:10:46-07:00, davidm @ tiger.hpl.hp.com [diffview]

Simplify the irq_desc[] patch to make it more acceptable for the
main tree.

ChangeSet 2003/06/19 12:15:44-07:00, davidm @ tiger.hpl.hp.com [diffview]

Fix weird PPC64 merge errors.

ChangeSet 2003/06/19 11:47:52-07:00, davidm @ tiger.hpl.hp.com [diffview]

defconfig:
  Update defconfig.

ChangeSet 2003/06/18 18:28:43-07:00, davidm @ tiger.hpl.hp.com [diffview]

Revert back to adjusting RPCSVC_MAXPAGES.

ChangeSet 2003/06/18 18:00:47-07:00, davidm @ tiger.hpl.hp.com [diffview]

Don't declare init_thread_union in sched.h if there is an architecture-
specific task-struct-allocator.

ChangeSet 2003/06/18 17:30:45-07:00, chadt @ sgi.com [diffview]

[PATCH] ia64: early_printk for SGI SN2

ChangeSet 2003/06/18 16:43:23-07:00, davidm @ tiger.hpl.hp.com [diffview]

More 2.5.72 cleanup/syncing.

ChangeSet 2003/06/17 23:57:02-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge conflicts with 2.5.72.

ChangeSet 2003/06/17 23:41:45-07:00, davidm @ tiger.hpl.hp.com [diffview]

Make hp-agp.c compile again.  It's still broken, because the generic
AGP code once again thinks all AGP bridges must be PCI devices.

ChangeSet 2003/06/17 23:38:10-07:00, davidm @ tiger.hpl.hp.com [diffview]

Add back /proc/sys/kernel/cache_decay_ticks.

ChangeSet 2003/06/17 23:33:39-07:00, davidm @ tiger.hpl.hp.com [diffview]

Fix <linux/pci.h> so it can be used on platforms with CONFIG_PCI off.

ChangeSet 2003/06/17 23:30:45-07:00, davidm @ tiger.hpl.hp.com [diffview]

Fix serial HCDP and ACPI support for serial driver.

ChangeSet 2003/06/16 18:55:43-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge with 2.5.71.

ChangeSet 2003/06/14 00:40:32-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/06/13 00:09:26-07:00, roland @ redhat.com [diffview]

[PATCH] make FIXMAP user page reading more portable

This patch vs 2.5.70 works fine for me on x86.  include/asm-x86_64/fixmap.h
needs the macros added for its vsyscall page too.

ChangeSet 2003/06/13 00:08:56-07:00, davidm @ tiger.hpl.hp.com [diffview]

memory.c:
  Undo last change.

ChangeSet 2003/06/12 19:25:55-07:00, davidm @ tiger.hpl.hp.com [diffview]

Allow read accesses to the entire FIXMAP range.  This lets gdb see the
instructions in the gate page (there is no security issue, because the
gate page only contains instructions or non-security-sensitive data such
as the ELF headers, etc.).

ChangeSet 2003/06/12 01:11:50-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/06/10 18:18:05-07:00, davidm @ tiger.hpl.hp.com [diffview]

Remove include of <asm/unistd.h> now that force_successful_syscall()
is back in ptrace.h.

ChangeSet 2003/06/10 18:14:41-07:00, davidm @ tiger.hpl.hp.com [diffview]

Fix gratuitous PPC64 changes (were probably caused by a bad bk merge
with Linus' tree).

ChangeSet 2003/06/10 13:44:20-07:00, davidm @ tiger.hpl.hp.com [diffview]

Clean up tg3 driver to avoid relying on PCI-DMA implementation internals.  This removes code and gets
rid of the (incorrect) dependence on PCI_DMA_BUS_IS_PHYS.

ChangeSet 2003/06/10 12:19:43-07:00, davidm @ tiger.hpl.hp.com [diffview]

Adjust for force_successful_syscall_return() having moved back to
ptrace.h.

ChangeSet 2003/06/10 01:01:20-07:00, davidm @ tiger.hpl.hp.com [diffview]

Don't try to support uncached accesses via read()/write().

ChangeSet 2003/06/06 22:57:36-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/06/05 14:00:20-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/06/05 13:52:04-07:00, davidm @ tiger.hpl.hp.com [diffview]

More small cleanups to /dev/mem driver and for stack dumping.

ChangeSet 2003/06/04 23:06:28-07:00, davidm @ tiger.hpl.hp.com [diffview]

Fix up drivers/char/mem.c some more.

ChangeSet 2003/06/04 22:58:44-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/06/04 18:45:48-07:00, davidm @ tiger.hpl.hp.com [diffview]

Clean up drivers/char/mem.c to reduce #ifdef bloat.  Make it work better on ia64.

ChangeSet 2003/06/04 14:14:42-07:00, davidm @ tiger.hpl.hp.com [diffview]

fork.c:
  More minor cleanups

ChangeSet 2003/06/04 13:56:27-07:00, davidm @ tiger.hpl.hp.com [diffview]

fork.c:
  Move task_cache back to its original place.

ChangeSet 2003/06/04 13:40:21-07:00, davidm @ tiger.hpl.hp.com [diffview]

Fix task creation/destruction to make it possible for the thread_info and the
task_struct to be next to each other.

ChangeSet 2003/06/03 23:42:21-07:00, davidm @ wailua.hpl.hp.com [diffview]

Add force_successful_syscall_return() to remaining platforms.

ChangeSet 2003/06/03 23:37:25-07:00, davidm @ wailua.hpl.hp.com [diffview]

Adjust show_trace()/show_stack() interfce to make them truly portable:
first arg is task whose trace should be shown, second arg is stack
pointer value starting from which the trace should be shown.  Either
(or both) arguments can be NULL.

ChangeSet 2003/06/03 09:28:48-07:00, davidm @ hpl.hp.com [diffview]

[PATCH] fix TCP roundtrip time update code

One of those very-hard-to-track-down, trivial-to-fix kind of problems:
without this patch, TCP roundtrip time measurements will corrupt the
routing cache's RTT estimates under heavy network load (the bug causes
RTAX_RTT to go negative, but since its type is u32, you end up with a
huge positive value...).  From there on, later TCP connections quickly
will go south.

The typo was introduced 8 months ago in v1.29 of the file by the patch
entitled "Cleanup DST metrics and abstrct MSS/PMTU further".

ChangeSet 2003/06/03 09:28:07-07:00, davidm @ tiger.hpl.hp.com [diffview]

ia64: Fix backtrace printing so it doesn't output tabs anymore.

ChangeSet 2003/05/30 17:39:18-07:00, peter @ chubb.wattle.id.au [diffview]

[PATCH] Fix ps/2 mouse and keyboard on I2000

I finally got sufficiently fed up to fix the PS/2 keyboard.  The
problem is that the generic code requests an ISA interrupt directly,
instead of via isa_irq_to_vector().

ChangeSet 2003/05/21 18:18:37-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/05/21 13:57:06-07:00, davidm @ tiger.hpl.hp.com [diffview]

Merge tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/linux-ia64-2.5

ChangeSet 2003/05/19 23:37:48-07:00, davidm @ tiger.hpl.hp.com [diffview]

Fix locking comments for time interpolator routines.

ChangeSet 2003/05/19 17:37:21-07:00, davidm @ tiger.hpl.hp.com [diffview]

Overhaul time-interpolation support.  Platforms that do not ever wish to
use time-interpolation should NOT define CONFIG_TIME_INTERPOLATION.
There will be no overhead on such platforms.

If you have a platform that has exactly one time-interpolation source,
define CONFIG_TIME_INTERPOLATION and let the generic code take care
of updating "last_nsec_offset" as needed, but never call
register_time_interpolator().  Instead, hard-code the updating of
last_nsec_offset in do_gettimeofday() (and do_settimeofday()).
This will minimize overhead to a single load & NULL check (everything else
gets inlined).

If you have a platform that needs to support multiple interpolation
sources, register each source with regsiter_time_interpolator().
Each call to time_interpolator_update()/time_interpolator_reset() will
then incur an indirect function call.  C'est la vie.

ChangeSet 2003/05/15 23:50:46-07:00, davidm @ tiger.hpl.hp.com [diffview]

Undo previous pasting-operator "fix".  Code was right and pre3.4 compiler was
(temporarily) broken.

ChangeSet 2003/05/14 18:04:06-07:00, davidm @ tiger.hpl.hp.com [diffview]

Add two hooks for high-precision time-interpolation.  Based on patch by
Jes Sorensen.

ChangeSet 2003/05/14 18:00:16-07:00, davidm @ tiger.hpl.hp.com [diffview]

Small driver fixes.

ChangeSet 2003/05/14 17:54:33-07:00, davidm @ tiger.hpl.hp.com [diffview]

Undo timer breakage.  Makes profil() work again.

ChangeSet 2003/05/14 00:55:04-07:00, davidm @ tiger.hpl.hp.com [diffview]

Change last_time_offset into last_nsec_offset and make it work for real.

ChangeSet 2003/05/14 00:53:14-07:00, davidm @ tiger.hpl.hp.com [diffview]

Misc fixes: PCI ids and prefetch in page-table tear-down.

ChangeSet 2003/05/14 00:49:39-07:00, davidm @ tiger.hpl.hp.com [diffview]

Add /proc/sys/kernel/cache_decay_ticks.

ChangeSet 2003/05/14 00:47:34-07:00, davidm @ tiger.hpl.hp.com [diffview]

Fix off-by-one error to make NFSd work with PAGE_SIZE = 64KB.

ChangeSet 2003/05/14 00:44:49-07:00, davidm @ tiger.hpl.hp.com [diffview]

Add MM_VM_SIZE() support.

ChangeSet 2003/05/14 00:42:39-07:00, davidm @ tiger.hpl.hp.com [diffview]

Add EARLY_PRINTK support.

ChangeSet 2003/05/14 00:40:25-07:00, davidm @ tiger.hpl.hp.com [diffview]

Minor build fixes/enhancements.

ChangeSet 2003/05/14 00:33:49-07:00, davidm @ tiger.hpl.hp.com [diffview]

select() speedup.

ChangeSet 2003/05/14 00:31:59-07:00, davidm @ tiger.hpl.hp.com [diffview]

Various minor driver fixes.

ChangeSet 2003/05/14 00:18:32-07:00, davidm @ tiger.hpl.hp.com [diffview]

Support serial ports in ACPI namespace and serial consoles via HCDP.

ChangeSet 2003/05/14 00:12:20-07:00, davidm @ tiger.hpl.hp.com [diffview]

Toolchain bug workarounds.

ChangeSet 2003/05/14 00:07:15-07:00, davidm @ tiger.hpl.hp.com [diffview]

AGP fixes for ia64.

ChangeSet 2003/05/14 00:04:26-07:00, davidm @ tiger.hpl.hp.com [diffview]

Make ACPI work on ia64.

ChangeSet 2003/05/14 00:01:23-07:00, davidm @ tiger.hpl.hp.com [diffview]

Generic DRM support for platforms that cant_use_aperture.

ChangeSet 2003/05/13 23:56:23-07:00, davidm @ tiger.hpl.hp.com [diffview]

Add module_arch_cleanup() callback.


Index generated at Sat Dec 20 08:39:39 UTC 2003