Simon Kelley [Mon, 18 Aug 2003 20:04:11 +0000 (16:04 -0400)]
[wireless atmel] minor updates
1) Add another card to the PCMCIA card database.
2) Fix a bug in wireless extensions.
3) Remove extra code for compilation without the firmware loader
4) force-enable CRC32 and FW_LOADER in Kconfig.
Andrew Morton [Mon, 18 Aug 2003 13:43:55 +0000 (06:43 -0700)]
[PATCH] cpumask_t: allow more than BITS_PER_LONG CPUs
From: William Lee Irwin III <wli@holomorphy.com>
Contributions from:
Jan Dittmer <jdittmer@sfhq.hn.org>
Arnd Bergmann <arnd@arndb.de>
"Bryan O'Sullivan" <bos@serpentine.com>
"David S. Miller" <davem@redhat.com>
Badari Pulavarty <pbadari@us.ibm.com>
"Martin J. Bligh" <mbligh@aracnet.com>
Zwane Mwaikambo <zwane@linuxpower.ca>
It has ben tested on x86, sparc64, x86_64, ia64 (I think), ppc and ppc64.
cpumask_t enables systems with NR_CPUS > BITS_PER_LONG to utilize all their
cpus by creating an abstract data type dedicated to representing cpu
bitmasks, similar to fd sets from userspace, and sweeping the appropriate
code to update callers to the access API. The fd set-like structure is
according to Linus' own suggestion; the macro calling convention to ambiguate
representations with minimal code impact is my own invention.
Specifically, a new set of inline functions for manipulating arbitrary-width
bitmaps is introduced with a relatively simple implementation, in tandem with
a new data type representing bitmaps of width NR_CPUS, cpumask_t, whose
accessor functions are defined in terms of the bitmap manipulation inlines.
This bitmap ADT found an additional use in i386 arch code handling sparse
physical APIC ID's, which was convenient to use in this case as the
accounting structure was required to be wider to accommodate the physids
consumed by larger numbers of cpus.
For the sake of simplicity and low code impact, these cpu bitmasks are passed
primarily by value; however, an additional set of accessors along with an
auxiliary data type with const call-by-reference semantics is provided to
address performance concerns raised in connection with very large systems,
such as SGI's larger models, where copying and call-by-value overhead would
be prohibitive. Few (if any) users of the call-by-reference API are
immediately introduced.
Also, in order to avoid calling convention overhead on architectures where
structures are required to be passed by value, NR_CPUS <= BITS_PER_LONG is
special-cased so that cpumask_t falls back to an unsigned long and the
accessors perform the usual bit twiddling on unsigned longs as opposed to
arrays thereof. Audits were done with the structure overhead in-place,
restoring this special-casing only afterward so as to ensure a more complete
API conversion while undergoing the majority of its end-user exposure in -mm.
More -mm's were shipped after its restoration to be sure that was tested,
too.
The immediate users of this functionality are Sun sparc64 systems, SGI mips64
and ia64 systems, and IBM ia32, ppc64, and s390 systems. Of these, only the
ppc64 machines needing the functionality have yet to be released; all others
have had systems requiring it for full functionality for at least 6 months,
and in some cases, since the initial Linux port to the affected architecture.
Albert Cahalan [Mon, 18 Aug 2003 05:32:22 +0000 (22:32 -0700)]
[PATCH] reduce diff between x86-64 & i386
This cleans up ioport.c to use BITS_PER_LONG, sizeof, and so on. This
makes it easier to spot the differences that matter, and thus easier to
find bugs.
David S. Miller [Sun, 17 Aug 2003 19:39:35 +0000 (12:39 -0700)]
[SCSI]: Fix bugs in sym2 hotplug conversion.
1) Make sym2_template.module get setup properly. It was
the scsi_module.c code taking care of this for us previously.
2) Zero out on-stack structures, in particular sym_dev not being
zeroed out was causing sometimes-boots-sometimes-doesnt problems
for me on sparc64 with sym2 non-modular.
Jamie Lokier [Sun, 17 Aug 2003 10:56:06 +0000 (03:56 -0700)]
[PATCH] make NFS lockd port numbers assignable at run time
When writing firewall rules, and you are serving NFS, it's really
useful to know the port numbers of the various NFS services. nfsd has
a standard value; mountd and statd are userspace daemons, and those
ports are settable on the command line.
The fiddly one is lockd. nlm_udpport and nlm_tcpport can be set on
the kernel command line or at module load time, but after that it's a
bit awkward (particularly as the lockd module can't be unloaded safely
- "rmmod -f lockd" sometimes panics).
This patch allows the port numbers and the other lockd parameters to
be set through files in /proc/sys/fs/nfs/nlm_*. The port numbers take
effect when lockd is next started or restarted.
In order to install the sysctl table even when compiled into the
kernel, it was necessary to update the initialisation code to the
current methods, using module_init() et al. This patch does that and
in so doing updates the module/kernel parameters to use the 2.6
module_param() method, as well as making the numeric range changes
consistent between the two.
Jamie Lokier [Sun, 17 Aug 2003 10:55:58 +0000 (03:55 -0700)]
[PATCH] use simple_strtoul for unsigned kernel parameters
The largest "unsigned int" value doesn't fit in a "long", on many machines.
So we should use simple_strtoul, not simple_strtol, to decode these values.
Adam Belay [Sun, 17 Aug 2003 10:24:02 +0000 (03:24 -0700)]
[PATCH] Fix sb_card.c for "name" removal
This will correct the sb_card compile error from removing "name" from
"struct device".
I also decided to read the name from the card structure instead of the
device structure because, for isapnp devices, the card structure is more
likely to contain an appropriate name.
Sam Ravnborg [Sun, 17 Aug 2003 09:29:08 +0000 (02:29 -0700)]
[PATCH] Move config tasks to kconfig/Makefile
This fixes a bug with multiple targets.
Olaf Hering reported that the build failed for PowerPc if used like
this: make oldconfig zImage
The reason for this was that .config was not present for any targets
specified in arch/$(ARCH)/Makefile and below.
That's because .config would not be included when oldconfig is present
in the list of targets. The fix is to move handling of *config task to
the kconfig/Makefile. Furthermore the logic in top-level makefile has
changed a bit, creating a more logial structure. When building a fresh
kernel, the user is now told that .config is missing, not an anonymous
report that .config did not exist.
The error has survided this long because the targets used in i386/boot
in general does not use CONFIG_ symbols.
James Bottomley [Sun, 17 Aug 2003 07:00:48 +0000 (02:00 -0500)]
scsi.h uses "u8" which isn't defined.
From: Muli Ben-Yehuda <mulix@mulix.org>
IMO, it's more correct to include <linux/types.h> in scsi.h, which
will bring in u8 and make scsi.h compilable on its own (provided
__KERNEL__ is defined, as it should be).
Jamie Lenehan [Sun, 17 Aug 2003 06:53:15 +0000 (01:53 -0500)]
[PATCH] dc395x - list handling cleanups
Here's another patch in the "try and make the driver readable"
cleanup series. This one cleans up all of the list handling.
Description: Replaced the various hand crafted scsi req lists
(waiting and going in the dcb and free in the acb) and the dcb list
(in the acb) with the generic linux lists. This makes all of the list
related code a *lot* more readable then it was previously.
Doug Ledford [Sun, 17 Aug 2003 09:29:33 +0000 (05:29 -0400)]
Reserve the sys_prctl() numbers for and add the stub for allowing
programs to select whether they use statistical time accounting
or accurate timestamp based accounting.
Jeff Garzik [Sun, 17 Aug 2003 08:23:28 +0000 (04:23 -0400)]
[arcnet com20020] misc fixes
* com20020_close expects two arguments (and actually uses the
second argument), but the arcnet layer only passes one arg.
Fun ensues.
* Remove __devinit markers, this is a library module.
* Move request_region up in com20020_found, to make the call
occur before the first I/O access in the function.
projects / linux / log