Linus Torvalds [Thu, 28 Feb 2002 04:45:11 +0000 (20:45 -0800)]
Fix off-by-one error reported by Anton Lavrentiev: we no
longer count the idle thread in "nr_threads", so we should
not discount it when returning sysinfo() information.
Alexander Viro [Thu, 28 Feb 2002 04:23:49 +0000 (20:23 -0800)]
[PATCH] (4/7) kdev_t removals
* write_some_buffers(), write_unlocked_buffers(), wait_for_buffers(),
wait_for_locked_buffers() and wait_for_some_buffers() converted
from kdev_t to struct block_device *.
Alexander Viro [Thu, 28 Feb 2002 04:23:44 +0000 (20:23 -0800)]
[PATCH] (3/7) kdev_t removals
* sync_buffers() split in two functions (sync_buffers() and
sync_all_buffers()). Callers of sync_buffers(NODEV) are
using the latter, those who actually pass a device - the former.
* sync_buffers() switched from kdev_t to struct block_device *.
Alexander Viro [Thu, 28 Feb 2002 04:23:33 +0000 (20:23 -0800)]
[PATCH] (1/7) kdev_t removals
* new function - fsync_bdev() (analog of fsync_dev(), but takes
struct block_device * instead of kdev_t. Callers of fsync_dev()
that have struct block_device in question are using fsync_bdev()
now.
* old code for fsync_dev(NODEV) had been moved to sys_sync().
Other callers of fsync_dev(NODEV) are calling sys_sync() now.
* fsync_dev() became a wrapper fro fsync_bdev().
* sync_dev() (not used anywhere in the tree) is gone.
* i2oblock.c had fsync_dev() called in ->release(). Removed.
* s390/block/xparm.c was doing fsync_dev() on its devices in
cleanup_module(). Removed.
Tim Waugh [Thu, 28 Feb 2002 04:19:29 +0000 (20:19 -0800)]
[PATCH] 2.5.6-pre1: parport and O_NONBLOCK
This patch makes lp and ppdev do the Right Thing regarding O_NONBLOCK.
2002-01-04 Tim Waugh <twaugh@redhat.com>
* include/linux/parport.h: Define a special inactivity timeout
meaning 'caller wants to use O_NONBLOCK'.
* drivers/char/lp.c: Support O_NONBLOCK properly.
* drivers/char/ppdev.c: Likewise.
* drivers/parport/parport_pc.c: Likewise.
* drivers/parport/ChangeLog: Updated.
Tim Waugh [Thu, 28 Feb 2002 04:19:24 +0000 (20:19 -0800)]
[PATCH] 2.5.6-pre1: shut lp up
This patch makes lp quieter in the common case that a printer does
_not_ speak ECP. (People have been writing to me worried that the
message means something bad.)
Martin Dalecki [Thu, 28 Feb 2002 04:15:37 +0000 (20:15 -0800)]
[PATCH] 2.5.6-pre1 IDE clean 14
Most importantly this patch is making ide.c use the
new automagic for module initialization lists and further
preparing the rest of the code in question here for proper
module separation. Despite this the CMOS probe has been removed
as well... *Iff*, which I don't expect, this breaks anything
it can be reintroduced easely. During this effort an actual bug
in the initialization of the main module has been uncovered as well.
a quite serious BUG has been tagged in ide-scsi.c as well, but
as far as now I just didn't get along to actually fixing it.
(The patch is big enough as it is).
Details follow:
- Kill *unused* ide_media_verbose() funciton.
- Remove the unnecessary media and supports_dma fields from
ide_driver_t.
- Remove the global name field from ide_driver_t struct by pushing it
down to the places where it's actually used.
- Remove the unused hwif_data field from ide_hwif_t.
- Push the supports_dsc_overlap condition up to the level where it
belongs: disk type as well.
- Make the initialization of ide main ide.c work with the new module
initialization auto-magic instead of calling it explicitly in
ll_rw_block.c This prevents the ide_init() from being called twice. We
have BTW. renamed it to ata_module_init(), since ata is more adequate
then ide and xxx_module_init corresponds better to the naming
conventions used elsewhere throughout the kernel.
This BUG was there before any ide-clean. It was worked around by a
magic variable preventing the second call to succeed. We have removed
this variable in one of the previous patches and thus uncovered it.
- Kill proc_ide_read_driver() and proc_ide_write_driver(). The drivers
already report on syslog which drives they have taken care of. (Or
at least they should). In esp. the proc_ide_write_driver() was just
too offending for me. Beleve it or not the purpose of it was to
*request a particular* driver for a device, by echoing some magic
values to a magic file...
More importantly this "back door" was getting in the way of a properly
done modularization of the IDE stuff.
- Made some not externally used functions static or not EXPORT-ed.
- Provide the start of a proper modularization between the main module
and drivers for particular device types. Changing the name-space
polluting DRIVER() macro to ata_ops() showed how inconsistently the
busy (read: module busy!) field from ide_driver_t
is currently used across the different device type modules.
This has to be fixed soon.
- Make the ide code use the similar device type ID numbers as the SCSI
code :-). This is just tedious, but it will help in a distant
feature. It helps reading the code anyway.
- Mark repettitive code with /* ATA-PATTERN */ comments for later
consolidation at places where we did came across it.
- Various comments and notes added where some explanations was missing.
Martin Dalecki [Thu, 28 Feb 2002 04:15:30 +0000 (20:15 -0800)]
[PATCH] 2.5.6-pre1 IDE clean 13a
This is finally moving the ide-pci.c file into a shape where
the host chip detection lists can finally be moved to where they
belong - into the particular chipset specific files.
This is accomplished, by a rather obivous removal of macro magic,
which was just making entries to the global device type
list nonfunctional, instead of making them conditional on the
corresponding CONFIG_BLHA options.
The second thing was to add a flag field to the device recognition
list, which made it possible to compress many of the
multi || chip id conditionals go away.
The only other file affected is ide.h - here is the change in the size
of the name field, which apparently slipped through ide-clean-12...
Martin Dalecki [Thu, 28 Feb 2002 04:15:25 +0000 (20:15 -0800)]
[PATCH] 2.5.5 IDE cleanup 12
1. Add some notes to Documentation/driver-model.txt about how and
and where to mount the driverfs.
2. Reorganize and prepare the PCI scanning code for proper device
dependant splitup. Basically tedious cleanup of macro games.
3. Use struct pci_dev name field as the name of PCI host dapaters
instead of invention ambigious IDE special names. This makes
the kernel bootup messages look a bit shifted, since those names are bit
longer, but makes up for consistance and should allow one later
to rearage things to fit into the generic PCI device initialization
mechanisms provided by the kernel.
4. Set 3. Allowed us to make the host chip specific
pci_init_xxx class functions have the proper signature of
module initializers. This will make it possible to make true
modules out of them later.
5. Make some functions in cmd64x.c static which where not used
elsewhere.
6. rename ide_special_settings to trust_pci_irq - this is reflecting
it's functionality better. And make it match the pci device vendor
as well as the device ID. It was a BUG to match only the device id!.
7. Make the chanell setup more tollerant for BIOS-es which don't
report IO and MEM bases properly. The code found previously there
tryed but was inconsistant.
8. Start to use proper terminology in ide-pci.c: host chip, channel,
drive instead of hwif, port, drive...
9. Enlarge the name field from ide_hwif_t to 64 bytes. It was only 6
previously and there where custom names there which where exceeding
this!!! But since we use the proper pci devce name there now instead,
we had to extend the size of this field anyway.
10. Add some explanatory comments and fix misguiding comments here and
there.
11. Kill the proc_ide_write_config and proc_ide_read_config brain
damage! Those where backdoors to the pci configuration registers on PCI
devices and IO registers on directly connected ISA ATA controllers.
They didn't discrement between them!
Access to both of them *simply* doesn't belong into an operating system,
which is supposed to abstract out the access to hardware! Did I mention
that access to both can be done from user land without an IDE special
interface! Any program which was using them (I hardly beleve there is
one) just deserves to loose. The programmer responsible for it
deserves to be fired immediately.
12. Move ide_map_xx and ide_unmap_xx tinny bio level wrappers away
from the "global" ide.h to where those are actually used and kill
trivial wrappers for otherwise generic bio_ routines. Just fighting
code obfuscation. The "rq->bio is used or is not there" brain
damage in ide-taskfile.c has to be fixed later. Possibly by killing
ide-taskfile.c alltogether, becouse this should be a driver for
users and not a driver for ATA disk disaster recovery companys...
13. Kill hwif->pci_devid and hwif->pci_venid. Just use the already
present hwif->pci_dev field instead.
14. Kill unused big switch ide_reinit_drive function. This silly
functon was switching upon every possible device driver cathegory
and calling the correspondng reinit function directly. This
idiocy was fortunately not used.
That's all... Most will be clear if one starts looking at the changes
in ide.h of course...
In contrast to the previous patches this one is actually fixing two
serious bugs.
The next direct step will be to kill the sigle place global PCI device
type recognition list from ide-pci.c by pushing the entries to where
they belong -> the host chips setup modules.
Martin Dalecki [Thu, 28 Feb 2002 04:15:18 +0000 (20:15 -0800)]
[PATCH] 2.5.6-pre1 IDE cleanup 11
1. Start of driver tree usage upon suggestion from Pavel Machek.
This still will needs a lot of further work in the future, but
the current code doesn't hurt anything and allowa Pavel to work
further from the base line. In esp. natively implemented
suspend to file requires this - which I would love to see comming
in,since I'm quite frequently using a notebook myself.
2. Kill the _IDE_C macro, which was playing games on entierly
unnecessary declarations inside of header files in esp ide_modes.h
3. Replace the functionally totally equal system_bus_block() and
ide_system_bus_speed() functions with one simple global
variable: system_bus_speed. This saves quite a significatn amount of
code. Unfortunately this is the part, which is makeing this
patch to appear bigger then it really is...
4. Use ide_devalidate_drive() directly instead of idedisk_revalidate().
5. Kill conditional CONFIG_KMOD as well as some other minor tweaks.
Well this isn't that much in terms of functionality, but it took me
quite q bit of time to catch up on the patch-2.5.5.gz ;-)
Martin Dalecki [Thu, 28 Feb 2002 04:15:12 +0000 (20:15 -0800)]
[PATCH] 2.5.6-pre1 IDE cleanup 10
This is finishing the cleanup parts already started in ide-clean-9.
It kills the ide_register_module() and ide_unregister_module() as well
as associated idiosyncracies alltogether. It turns out
that this patch is actually fixing a bug which was present in the
driver before: the sub-module initialization functions where called
at least twice - which is an abundance.
Tough there is a bit of global namespace pollution caused by this
patch - but I'm aware of it and will fix it just a bit later.
(The terminology used inside the IDE code is anyway nothing common
else in the linux universum...)
The next targets will be:
1. Code obfuscation by "wrappers" around generic BIO level functions.
2. ide_hwgroup_t - which is only used to serialize multiple
discs on the same interrupt and similar. This is however a tough one.
3. There is a plenty of code waste in the chipset drivers, where there
is baroque informative code for the proc file system for static stuff,
which in fact belongs just to syslog(). In fact the default RedHat
distribution kernel is killing this gratitious abuse of the /proc
concept since a long long time...
I'm still awaiting the day of /proc/GPL, where GPL contains the
full text of it...
Linus Torvalds [Thu, 28 Feb 2002 04:11:56 +0000 (20:11 -0800)]
Shrink icache more aggressively - if we free the dentries,
go ahead and free the inodes too, don't try to age them any
more (the aging has been done on a dentry level).
Steven Cole [Wed, 27 Feb 2002 14:20:26 +0000 (09:20 -0500)]
This patch adds help texts for CONFIG_SERIAL_TX3912,
CONFIG_SERIAL_TX3912_CONSOLE, CONFIG_AU1000_SERIAL_CONSOLE,
CONFIG_AU1000_UART, CONFIG_EUROTECH_WDT to drivers/char/Config.help.
usb visor driver:
- reworked urb handling, getting rid of lots of code now that we have
proper urb reference counting.
- removed port locks as the usb-serial core now does this.
- added support for the Palm m515 thanks to SilaS
usb serial core:
- cleaned up some whitespace issues
- changed MOD_INC logic for the generic driver
- the port->sem lock is now taken by the serial core, not the individual
usb-serial drivers. This is to reduce races.
Johannes Erdfelt [Wed, 27 Feb 2002 07:16:43 +0000 (23:16 -0800)]
uhci.c, speed improvements
Basically, the patch turns switching off FSBR into a lazy operation with
the assumption there will be another transfer shortly afterwards. This
works wonders for usb-storage for instance.
David Brownell [Wed, 27 Feb 2002 07:15:11 +0000 (23:15 -0800)]
This is another USB API cleanup patch. It's against 2.5.5:
- Moves 8 functions from usb.[hc] to hcd.[hc]
- Also moves some data structures and types
- Now usbdevfs and "old" HCDs #include "hcd.h"
- Minor tweaks to the "hcd" layer (one less FIXME)
- Minor kernel doc and comment cleanups
Basically this continues moving the HCD-only functionality
out of the way of normal USB device drivers. Converging
"usb_bus" and "usb_hcd" (later!) will be a bit easier too.
I did basic sanity tests, there's little to break ... :)
There are still a few functions in usb.c that aren't for
general driver use. They're mostly for enumeration,
in areas where the hub driver and HCD root hubs
need to do various kinds of magic. It wasn't clear
how to decouple those, they can certainly wait.
David Brownell [Wed, 27 Feb 2002 07:13:10 +0000 (23:13 -0800)]
PATCH: audio driver urb allocation
This fixes a bug in the audio driver which came from an
incorrect conversion from static to dynamic URB allocation.
It's against 2.5.5
I noticed this while trying to see exactly how ISO transfers
get used. The bug is that while originally the driver statically
allocated several structures {urb + N * iso packet descriptors},
the update forgot to allocate the ISO descriptors.
Likely not many folk noticed this on 32 bit machines, where
sizeof urb == 92, because kmalloc rounds that up to 128,
adding 36 bytes of external padding. The ISO descriptors
took up 32 bytes of that, which "just happened" to already
have been allocated but unused.
David Brownell [Wed, 27 Feb 2002 07:09:12 +0000 (23:09 -0800)]
[PATCH] hid-core and hotplug
This restores a line someone deleted, which
affects hotplugging. Basically this restores
correct/previous behavior: the HID driver only
matches HID devices, not every device that
ever connects.
USB auerswald driver:
- changed the minor number the auerswald driver was using, as it was found
out that this number was already in use by another USB driver!
PCI Hotplug Core cleanups:
- pcihpfs cleanup, removing unneeded file operations.
- Added facility to have the files change their timestamps if the data
within the file changes.
Dave Jones [Tue, 26 Feb 2002 08:57:44 +0000 (00:57 -0800)]
[PATCH] updates.
Forward ports from 2.4, Various janitor bits, and some fixes by me to
make the thing work again in 2.5. I munged the MTDRAM driver to work
also (seperate patch to follow), and it seems to work.
David Woodhouse gave this the once over, and approved the changes.
Complete changelog below:
o Don't create two slabcaches with the same name.
o Don't corrupt eraseblock lists on mount
o Don't mark nodes obsolete during mount
o __attribute__((packed)) on the node definitions.
o Fix up() without down() in jffs2_readdir().
o Fix duplicate version number usage - s/highest_version++/++highest_version/
o Fix (i.e. implement) mtime/ctime on directories.
maybe too busy with the bk stuff
o Don't allow hardlinks of directories.
o s/(mode&S_IFMT)==S_IFLNK/S_ISLNK(mode)/ et al to keep Al happy.
o Fix for garbage-collection of holes, where we used to write nodes out
with csize/dsize swapped. Workarounds for existing such brokenness.
o Improve wear levelling by rotating node lists on mount, to avoid starting
at one end of the flash every time.
o Remember to get internal inode-semaphore on symlink operations.
Neil Brown [Tue, 26 Feb 2002 06:23:49 +0000 (22:23 -0800)]
[PATCH] PATCH 13/16: NFSD: TCP: Reserve space on sndbuf so we never block when writing
Make sure there is alway adequate sndbuf space for replies.
We keep track of how much space might be needed for replies
and never dequeue a request unless there is adequate space
for a maximal reply. We assume each request will generate a maximal
sized reply until the request is partly decoded.
Each RPC program/procedure can specify the maximum size
of a reply to the precedure (though they don't yet).
The wspace callback is used to enqueue sockets that may be waiting
for sndbuf space to become available.
As there should always be enough buffer space to the full
reply, the only reason that sock_sendmsg could block is due
to a kmalloc delay. As this is likely to be fairly quick (and if
it isn't the server is clagged anyway) we remove the MSG_DONTWAIT
flag, but set a 30 second timeout on waiting. If the wait
ever times out, we close the connection. If it doesn't we can
be sure that we did a complete write.
When a request completes, we make sure that the space
used for the reply does not exceed the space reserved. This
is an internal consistancy check.
This patchs sets the sndbuf and rcvbuf sizes for all sockets
used for rpc service. This size if dependant on the servers bufsize (S) and
partially on the number of threads (N).
For UDP
sndbuf == 5*S
rcvbuf == (N+2)*S
for TCP
sndbuf == N*S
rcvbuf == 3*S
Neil Brown [Tue, 26 Feb 2002 06:23:44 +0000 (22:23 -0800)]
[PATCH] PATCH 12/16: NFSD: TCP: Close idle TCP connections
Close idle rpc/tcp sockets
We split the list of sv_allsocks into two, one
of permanent sockets (udp, tcp listener) and one
of temporary sockets (tcp data).
Whenever we complete a successful receive on a temp socket,
it gets pushed to the end of the list.
Whenever a thread wants to do something, it first checks if
the oldest temp socket has not has a receive for 6 mintutes
(should possibly be configurable). It so, we simulate
a close.
Finally we make sure that threads wake up every few minutes
so that if the server is completely idle, all temp
sockets will get closed.
Neil Brown [Tue, 26 Feb 2002 06:23:37 +0000 (22:23 -0800)]
[PATCH] PATCH 11/16: NFSD: TCP: close bad connections
Detect and close tcp connections that we cannot work with.
If an rpc fragment that arrives on a tcp connection
is non-terminal or too large for our buffer, then we
have to close the connection.
Also, if a write fails on a tcp connection, we close
the connection.
Neil Brown [Tue, 26 Feb 2002 06:23:32 +0000 (22:23 -0800)]
[PATCH] PATCH 10/16: NFSD: TCP: rationalise locking in RPC server routines
Tidy up SMP locking for svc_sock
sk_lock is not necessary and is now removed.
The only things that were happening under sk_lock but
not the more global sv_lock were testing and setting
some of the flags: sk_busy, sk_conn, sk_data etc.
These have been changed to bits in a flags word which are atomically
set and tested.
Also, by establishing some simple rules about that must
be done after setting these flags, the locking is not needed.
With this patch sk_conn and sk_data are now flags, not counts (sk_data
was already a flag for udp). They are set if there might be
a connection or data, and only clear when we are sure there aren't
(or when we are about to check if there is).
svc_sock_accepted becomes identical to svc_sock_recieved and
so is discarded in favour of the latter.
Neil Brown [Tue, 26 Feb 2002 06:23:22 +0000 (22:23 -0800)]
[PATCH] PATCH 8/16: NFSD: RPC lists tidyup
Change sunrpc to use more list.h lists
The sunrpc client code uses home-grown doubly linked
lists to group
- idle server threads
- pending server sockets
- waiting rpc tasks
- all rpc tasks.
This patch converts all of these lists to <linux/list.h> lists and
also makes the list of all server sockets for a particular server into
a list.h list instead of a single-link list.
Possibly the least obvious change is replacing RPC_INIT_WAITQ
with RPC_WAITQ and INIT_RPC_WAITQ. These follow the model of
LIST_HEAD and INIT_LIST_HEAD defined in list.h and are needed
to initialise the list_head in the rpc_waitq properly.
Remove BKL from nfsservctl systemcall. All
components have their own locking.
Also remove it from the body of nfsd threads. Keep
it for final thread destruction though.
Neil Brown [Tue, 26 Feb 2002 06:23:05 +0000 (22:23 -0800)]
[PATCH] PATCH 5/16: NFSD: BKL removal: add BKL where needed in filehandle lookup
Protect dentry attachement from BKL
The process of attaching a dentry into the dcache
still needs the BKL I think.
When all the other BKL changes in the VFS settle down, I
will revisit this. But as it is not a very frequent
operation, the BKL wont hurt.
Also add a down/up of i_sem when doing a lookup(".."),
as it is down for all other lookups.
Neil Brown [Tue, 26 Feb 2002 06:23:00 +0000 (22:23 -0800)]
[PATCH] PATCH 4/16: NFSD: BKL Removal: Locking in nfssvc.c
Tidy up locking in nfssvc - preparing for BKL removal
- nfsd_busy becomes atomic_t
- nfsd_call_lock created to protect timing of calls and stats gathering
- lock_kernel around thread creation and destruction. It is
sufficiently uncommon that it doesn't really need a lock of it's
own. It is currently under the BKL because all of the nfsservctl
syscall is, but that is about to be removed so we preserve BKL here.
Change export table lock to (SMP safe) rwsemaphore
As a first step to removing the BKL from nfsd, this patch
changes the lock used for the export table to be a rwsem semaphore.
Previously it had the same functionality but depended on the BKL
for correctness.
As there is no "down_write_interruptible" this patch removes the
posibility of interrupting the write_lock request, but this should
never be needed anyway.
Linus Torvalds [Tue, 26 Feb 2002 05:27:08 +0000 (21:27 -0800)]
Merge penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/joystick
into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux
Vojtech Pavlik [Tue, 26 Feb 2002 05:26:19 +0000 (21:26 -0800)]
[PATCH] Input patch - joystick drivers update
As a step in the process of updating all the linux input drivers to the
most recent version available, this patch updates all the joystick
drivers that are in the kernel tree.
Changes are mainly the input hotplug/proc support in every of the
joystick drivers, and fixing trivial bugs here and there. I can supply a
more detailed description upon request.
(The addition of the parentheses is needed so the shift will take
place after the bitwise-or. Without the parentheses, the shift will
incorrectly always take place before the bitwise-or.)
Andi Kleen [Tue, 26 Feb 2002 05:19:05 +0000 (21:19 -0800)]
[PATCH] x86_64 irqstat change
x86_64 does not have an irq_stat array. It puts the interrupt information
into its per CPU area instead, allowing more efficient access to it.
This patch allows the architecture to overwrite
the __IRQ_STAT access macro in linux/irq_cpustat.h.
Andi Kleen [Tue, 26 Feb 2002 05:16:06 +0000 (21:16 -0800)]
[PATCH] x86-64 update for 2.5.5
This patch makes x86-64 compile in 2.5.5 and syncs it with changes in the i386
port. It also fixes some bugs that were discovered in recent testing:
- enhance 32bit emulation and fix bugs.
- fix security hole in vmalloc handling
- Do not use lockless gettimeofday for now because it is buggy.
The patch only changes x86_64 specific files.
Jaroslav Kysela [Tue, 26 Feb 2002 03:47:57 +0000 (19:47 -0800)]
[PATCH] ALSA patch for 2.5.5
- added support for KERN_ prefixes for snd_printk
- verbose printk (file:number) is now optional
- fixed devfs OSS device names in soundcore
- added XRUN ioctl for PCM API
- improved support for Sound Blaster Audigy
- fixed AC3 forwarding for Sound Blaster Live!
- more fixes in dependencies in Makefiles
- llseek locking fixes for sound/core/info.c
- fixed ISA DMA allocation
- added wait() callback for ac97_codec.c and VIA686 driver
- CMIPCI driver updates
- added AMD-8111 support for Intel8x0
Dave Jones [Tue, 26 Feb 2002 03:36:01 +0000 (19:36 -0800)]
[PATCH] later DMI scan.
This moves the dmi scan to an earlier stage so that we can trap issues
such as the various laptops that don't like enabling APIC.
It's likely to be useful for trapping other similar early-boot problems.
Originally by Mikael Pettersson
Dave Jones [Tue, 26 Feb 2002 03:35:39 +0000 (19:35 -0800)]
[PATCH] un'fix' NCR scsi driver.
Linus,
Current driver in your tree has been 'fixed' by someone without
understanding just how broken it was. Numerous fixes were done in 2.4
after the 2.5 split by Alan.
This patch reverts the damage the driver does in your tree, and brings
Alan's 2.4 bits over instead. Downside: It doesn't compile.
Upside: It doesn't pretend to work and corrupt data, and at least
is the right base for people to start fixing.
Russell King [Mon, 25 Feb 2002 19:43:27 +0000 (19:43 +0000)]
Rationalise interrupt handling on ARM. With the old code, we had
loops within loops within loops checking until each IRQ level didn't
have any more interrupts for us. This caused both latency problems,
and locked out any chance of handling a second interrupt from down
the chain while one on that chain was already in progress.
The new structure splits out the machine specific IRQ handlers from
the Linux driver specific IRQ handlers, giving the machine specific
handlers much greater flexibility in handling the interrupt. We
also suck the SA1100 IRQ edge selection function into the IRQ core.
Russell King [Mon, 25 Feb 2002 11:07:23 +0000 (11:07 +0000)]
Fix nwfpe so GDB can debug user space floating point again.
Patch 960/1 (Peter Teichmann):
NWFPE patch to be more compliant to IEEE-754
1. The RND/URD instruction was handled as int_to_float(float_to_int
(number)) which is wrong because it only works for floating point
numbers that fit in an integer.
2. The FLT instruction was setting the rounding precision for
extended precision calculations, which is not necessary
(probably a historic relict) but has undesirable side effects
on all extended precision calculations.
Russell King [Mon, 25 Feb 2002 10:24:56 +0000 (10:24 +0000)]
Clean up ARM TLB handling code; previously there was a lot of code
replication across each processor type, each handling alignment of
addresses slightly differently. We unify this mess, and allow for
greater flexibility in the per-CPU architecture TLB handlers.
We also start to remove the ARM cache.h -> cpu_*.h -> proc-fns.h mess
making the code cleaner and easier to follow.
Documentation describing the expected behaviour of each TLB function
for the 32-bit ARM processors is also included.