Pierre Ossman [Thu, 27 Sep 2007 08:48:29 +0000 (10:48 +0200)]
sdio: adaptive interrupt polling
The interrupt polling frequency is a compromise between power usage and
interrupt latency. Unfortunately, it affects throughput rather severely
for devices which require an interrupt for every chunk of data.
By making the polling frequency adaptive, we get better throughput with
those devices without sacficing too much power. Polling will quickly
increase when there is an actual interrupt, and slowly fall back to the
idle frequency when the interrupts stop coming.
David Brownell [Wed, 8 Aug 2007 16:12:54 +0000 (09:12 -0700)]
mmc_spi host driver
This is the latest version of the MMC-over-SPI support. It works
on 2.6.23-rc2 plus git-mmc (from rc1-mm2), along with the preceding
patches which teach the rest of the MMC stack about SPI.
The main issue of note is that sometimes cards need to be power cycled
to recover after certain faults. Also, it may sometimes be necessary
to disable CRCs. ("modprobe mmc_core use_spi_crc=n")
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: mikael.starvik@axis.com, Cc: Hans-Peter Nilsson <hp@axis.com> Cc: Jan Nikitenko <jan.nikitenko@gmail.com> Cc: Mike Lavender <mike@steroidmicros.com> Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
David Brownell [Wed, 8 Aug 2007 16:11:32 +0000 (09:11 -0700)]
MMC core learns about SPI
Teach the MMC/SD/SDIO core about using SPI mode.
- Use mmc_host_is_spi() so enumeration works through SPI signaling
and protocols, not just the native versions.
- Provide the SPI response type flags with each request issued,
including requests from the new lock/unlock code.
- Understand that cmd->resp[0] and mmc_get_status() results for SPI
return different values than for "native" MMC/SD protocol; this
affects resetting, checking card lock status, and some others.
- Understand that some commands act a bit differently ... notably:
* OP_COND command doesn't return the OCR
* APP_CMD status doesn't have an R1_APP_CMD analogue
Those changes required some new and updated primitives:
- Provide utilities to access two SPI-only requests, and one
request that wasn't previously needed:
* mmc_spi_read_ocr() ... SPI only
* mmc_spi_set_crc() ... SPI only (override by module parm)
* mmc_send_cid() ... for use without broadcast mode
- Updated internal routines:
* Previous mmc_send_csd() modified into mmc_send_cxd_native();
it uses native "R2" responses, which include 16 bytes of data.
* Previous mmc_send_ext_csd() becomes new mmc_send_cxd_data()
helper for command-and-data access
* Bugfix to that mmc_send_cxd_data() code: dma-to-stack is
unsafe/nonportable, so kmalloc a bounce buffer instead.
- Modified mmc_send_ext_csd() now uses mmc_send_cxd_data() helper
- Modified mmc_send_csd(), and new mmc_spi_send_cid(), routines use
those helper routines based on whether they're native or SPI
The newest categories of cards supported by the MMC stack aren't expected
to work yet with SPI: MMC or SD cards with over 4GB data, and SDIO.
All those cards support SPI mode, so eventually they should work too.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
Nicolas Pitre [Mon, 20 Aug 2007 21:17:37 +0000 (17:17 -0400)]
sdio: fix recursion issues between sdio-uart driver and tty layer
In a few places, sdio_uart_irq() is called directly instead of waiting
for the actual interrupt to be raised and the SDIO IRQ thread scheduled
in order to reduce latency. However, some interaction with the tty core
may end up calling us back (serial echo, flow control, etc.) creating
two issues:
- the host lock gets claimed twice from the same thread causing a
deadlock;
- the same direct calls to sdio_uart_irq() may be performed causing
unexpected reentrancy into the IRQ handler.
This patch handles both of those issues.
Signed-off-by: Nicolas Pitre <npitre@mvista.com> Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
Nicolas Pitre [Wed, 15 Aug 2007 17:27:29 +0000 (13:27 -0400)]
sdio: add default c_ispeed/c_ospeed values to sdio_uart driver
Note that the default baudrate is 4800 instead of 9600 as a convenience
because that's what GPS devices want which is still the main use for
this driver.
Signed-off-by: Nicolas Pitre <npitre@mvista.com> Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
David Vrabel [Fri, 10 Aug 2007 12:29:46 +0000 (13:29 +0100)]
sdio: add sdio_f0_readb() and sdio_f0_writeb()
Add sdio_f0_readb() and sdio_f0_writeb() functions to reading and
writing function 0 registers. Writes outside the vendor specific CCCR
registers (0xF0 - 0xFF) are not permitted.
Signed-off-by: David Vrabel <david.vrabel@csr.com> Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
David Vrabel [Wed, 8 Aug 2007 13:24:21 +0000 (14:24 +0100)]
sdio: extend sdio_readsb() and friends to handle any length of buffer
Extend sdio_readsb(), sdio_writesb(), sdio_memcpy_fromio(), and
sdio_memcpy_toio() to handle any length of buffer by splitting the transfer
into several IO_RW_EXTENDED commands. Typically, a transfer would be split
into a single block mode transfer followed by a byte mode transfer for the
remainder but we also handle lack of block mode support and the block size
being greater than 512 (the maximum byte mode transfer size).
host->max_seg_size <= host->max_req_size so there's no need to check both
when determining the maximum data size for a single command.
Signed-off-by: David Vrabel <david.vrabel@csr.com> Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
David Vrabel [Wed, 8 Aug 2007 13:23:48 +0000 (14:23 +0100)]
sdio: set the functions' block size
Before a driver is probed, set the function's block size to the default so the
driver is sure the block size is something sensible and it needn't explicitly
set it.
The default block size is the largest that's supported by both the card and
the host, with a maximum of 512 to ensure aribitrarily sized transfer use the
optimal (least) number of commands.
See http://lkml.org/lkml/2007/8/7/150 for reasons for the block size choice.
Signed-off-by: David Vrabel <david.vrabel@csr.com> Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
Nicolas Pitre [Sat, 16 Jun 2007 06:07:53 +0000 (02:07 -0400)]
mmc: initialize mmc subsystem with subsys_initcall()
The problem is that the sdio_bus must be registered before any SDIO
drivers are registered against it otherwise the kernel sulks. Because
the sdio_bus registration happens through module_init (equivalent to
device_initcall), then any SDIO
drivers linked before the SDIO core code in the kernel will be initialized
first.
Upcoming SDIO function drivers are likely to be located outside the
drivers/mmc directory as it is common practice to group drivers according
to their function rather than the bus they use. SDIO drivers are therefore
likely to appear at random location in the kernel link.
To make sure the sdio_bus is always initialized before any SDIO drivers,
let's move the MMC init to the subsys_initcall level.
Signed-off-by: Nicolas Pitre <npitre@mvista.com> Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
Pierre Ossman [Mon, 30 Jul 2007 13:15:30 +0000 (15:15 +0200)]
sdio: split up common and function CIS parsing
Add a more clean separation between global, common CIS information
and the function specific one as we need the common information in
places where no specific function is specified.
Nicolas Pitre [Sat, 16 Jun 2007 06:06:47 +0000 (02:06 -0400)]
sdio: link unknown CIS tuples to the sdio_func structure
This way those tuples that the core cares about are consumed by the core
code, and tuples that only function drivers might make sense of are
available to drivers.
Signed-off-by: Nicolas Pitre <npitre@mvista.com> Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
Thomas Gleixner [Sat, 22 Sep 2007 22:29:06 +0000 (22:29 +0000)]
clockevents: remove the suspend/resume workaround^Wthinko
In a desparate attempt to fix the suspend/resume problem on Andrews
VAIO I added a workaround which enforced the broadcast of the oneshot
timer on resume. This was actually resolving the problem on the VAIO
but was just a stupid workaround, which was not tackling the root
cause: the assignement of lower idle C-States in the ACPI processor_idle
code. The cpuidle patches, which utilize the dynamic tick feature and
go faster into deeper C-states exposed the problem again. The correct
solution is the previous patch, which prevents lower C-states across
the suspend/resume.
Remove the enforcement code, including the conditional broadcast timer
arming, which helped to pamper over the real problem for quite a time.
The oneshot broadcast flag for the cpu, which runs the resume code can
never be set at the time when this code is executed. It only gets set,
when the CPU is entering a lower idle C-State.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Andrew Morton <akpm@linux-foundation.org> Cc: Len Brown <lenb@kernel.org> Cc: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com> Cc: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Thomas Gleixner [Sat, 22 Sep 2007 22:29:05 +0000 (22:29 +0000)]
ACPI: disable lower idle C-states across suspend/resume
device_suspend() calls ACPI suspend functions, which seems to have undesired
side effects on lower idle C-states. It took me some time to realize that
especially the VAIO BIOSes (both Andrews jinxed UP and my elfstruck SMP one)
show this effect. I'm quite sure that other bug reports against suspend/resume
about turning the system into a brick have the same root cause.
After fishing in the dark for quite some time, I realized that removing the ACPI
processor module before suspend (this removes the lower C-state functionality)
made the problem disappear. Interestingly enough the propability of having a
bricked box is influenced by various factors (interrupts, size of the ram image,
...). Even adding a bunch of printks in the wrong places made the problem go
away. The previous periodic tick implementation simply pampered over the
problem, which explains why the dyntick / clockevents changes made this more
prominent.
We avoid complex functionality during the boot process and we have to do the
same during suspend/resume. It is a similar scenario and equaly fragile.
Add suspend / resume functions to the ACPI processor code and disable the lower
idle C-states across suspend/resume. Fall back to the default idle
implementation (halt) instead.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Andrew Morton <akpm@linux-foundation.org> Cc: Len Brown <lenb@kernel.org> Cc: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com> Cc: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Merge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6
* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6:
ACPI: suspend: consolidate handling of Sx states addendum
ACPI: suspend: consolidate handling of Sx states.
ACPI: video: remove dmesg spam
ACPI: video: _DOS=0 by default to prevent hotkey hang
Frans Pop [Thu, 20 Sep 2007 20:27:44 +0000 (22:27 +0200)]
ACPI: suspend: consolidate handling of Sx states addendum
Make the S0 state be always reported as supported
Signed-off: Frans Pop <elendil@planet.nl> Acked-by: Alexey Starikovskiy <astarikovskiy@suse.de> Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Len Brown <len.brown@intel.com>
Suresh Siddha points out that this one breaks the fundamental
requirement that you cannot free page table pages before the TLB caches
are flushed. The quicklists do not give the same kinds of guarantees
that the mmu_gather structure does, at least not in NUMA configurations.
Requested-by: Suresh Siddha <suresh.b.siddha@intel.com> Acked-by: Andi Kleen <ak@suse.de> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Christoph Lameter <clameter@sgi.com> Cc: Asit Mallick <asit.k.mallick@intel.com> Cc: Tony Luck <tony.luck@intel.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Merge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2
* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2:
ocfs2: Pack vote message and response structures
ocfs2: Don't double set write parameters
ocfs2: Fix pos/len passed to ocfs2_write_cluster
ocfs2: Allow smaller allocations during large writes
Recent changes to sleep initialization in ACPI dropped reporting of supported Sx
states above S3. Fix that and also move S5 init into same file as other Sx.
The only functional change is adding printk() for S4 and S5 cases.
Signed-off-by: Alexey Starikovskiy <astarikovskiy@suse.de> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Len Brown <len.brown@intel.com>
The ocfs2_vote_msg and ocfs2_response_msg structs needed to be
packed to ensure similar sizeofs in 32-bit and 64-bit arches. Without this,
we had inadvertantly broken 32/64 bit cross mounts.
Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com> Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Mark Fasheh [Wed, 19 Sep 2007 00:49:29 +0000 (17:49 -0700)]
ocfs2: Don't double set write parameters
The target page offsets were being incorrectly set a second time in
ocfs2_prepare_page_for_write(), which was causing problems on a 16k page
size kernel. Additionally, ocfs2_write_failure() was incorrectly using those
parameters instead of the parameters for the individual page being cleaned
up.
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Mark Fasheh [Mon, 17 Sep 2007 16:06:29 +0000 (09:06 -0700)]
ocfs2: Fix pos/len passed to ocfs2_write_cluster
This was broken for file systems whose cluster size is greater than page
size. Pos needs to be incremented as we loop through the descriptors, and
len needs to be capped to the size of a single cluster.
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Mark Fasheh [Mon, 17 Sep 2007 03:10:16 +0000 (20:10 -0700)]
ocfs2: Allow smaller allocations during large writes
The ocfs2 write code loops through a page much like the block code, except
that ocfs2 allocation units can be any size, including larger than page
size. Typically it's equal to or larger than page size - most kernels run 4k
pages, the minimum ocfs2 allocation (cluster) size.
Some changes introduced during 2.6.23 changed the way writes to pages are
handled, and inadvertantly broke support for > 4k page size. Instead of just
writing one cluster at a time, we now handle the whole page in one pass.
This means that multiple (small) seperate allocations might happen in the
same pass. The allocation code howver typically optimizes by getting the
maximum which was reserved. This triggered a BUG_ON in the extend code where
it'd ask for a single bit (for one part of a > 4k page) and get back more
than it asked for.
Fix this by providing a variant of the high level allocation function which
allows the caller to specify a maximum. The traditional function remains and
just calls the new one with a maximum determined from the initial
reservation.
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Merge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev
* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev:
[libata] ahci: add ATI SB800 PCI IDs
libata-sff: Fix documentation
libata: Update the blacklist with a few more devices
This simplifies signalfd code, by avoiding it to remain attached to the
sighand during its lifetime.
In this way, the signalfd remain attached to the sighand only during
poll(2) (and select and epoll) and read(2). This also allows to remove
all the custom "tsk == current" checks in kernel/signal.c, since
dequeue_signal() will only be called by "current".
I think this is also what Ben was suggesting time ago.
The external effect of this, is that a thread can extract only its own
private signals and the group ones. I think this is an acceptable
behaviour, in that those are the signals the thread would be able to
fetch w/out signalfd.
Wolfgang Walter [Thu, 20 Sep 2007 19:51:46 +0000 (15:51 -0400)]
rpc: fix garbage in printk in svc_tcp_accept()
we upgraded the kernel of a nfs-server from 2.6.17.11 to 2.6.22.6. Since
then we get the message
lockd: too many open TCP sockets, consider increasing the number of nfsd threads
lockd: last TCP connect from ^\\236^\É^D
These random characters in the second line are caused by a bug in
svc_tcp_accept.
(Note: there are two previous __svc_print_addr(sin, buf, sizeof(buf))
calls in this function, either of which would initialize buf correctly;
but both are inside "if"'s and are not necessarily executed. This is
less obvious in the second case, which is inside a dprintk(), which is a
macro which expands to an if statement.)
Signed-off-by: Wolfgang Walter <wolfgang.walter@studentenwerk.mhn.de> Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
A driver writer from another operating system hinted that
the versions of Yukon 2 chip with rambuffer (EC and XL) have
a hardware bug that if the FIFO ever gets completely full it
will hang. Sounds like a classic ring full vs ring empty wrap around
bug.
As a workaround, use the existing watchdog timer to check for
ring full lockup.
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
This patch should cause no functional changes in driver behaviour.
There are (too) many revisions of the Yukon 2 chip now. Instead of
adding more conditionals based on chip revision; rerganize into a
set of feature flags so adding new versions is less problematic.
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
The length check for truncated frames was not correctly handling
the case where VLAN acceleration had already read the tag.
Also, the Yukon EX has some features that use high bit of status
as security tag.
Signed-off-by: Pierre-Yves Ritschard <pyr@spootnik.org> Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
Stefan Richter [Thu, 20 Sep 2007 19:17:33 +0000 (21:17 +0200)]
ieee1394: ohci1394: fix initialization if built non-modular
Initialization of ohci1394 was broken according to one reporter if the
driver was statically linked, i.e. not built as loadable module. Dmesg:
PCI: Device 0000:02:07.0 not available because of resource collisions
ohci1394: Failed to enable OHCI hardware.
This was reported for a Toshiba Satellite 5100-503. The cause is commit 8df4083c5291b3647e0381d3c69ab2196f5dd3b7 in Linux 2.6.19-rc1 which only
served purposes of early remote debugging via FireWire. This
functionality is better provided by the currently out-of-tree driver
ohci1394_earlyinit. Reversal of the commit was OK'd by Andi Kleen.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Herbert Xu [Wed, 19 Sep 2007 17:46:28 +0000 (10:46 -0700)]
[PPP] L2TP: Fix skb handling in pppol2tp_xmit
This patch makes pppol2tp_xmit call skb_cow_head so that we don't modify
cloned skb data. It also gets rid of skb2 we only need to preserve the
original skb for congestion notification, which is only applicable for
ppp_async and ppp_sync.
The other semantic change made here is the removal of socket accounting
for data tranmitted out of pppol2tp_xmit. The original code leaked any
existing socket skb accounting. We could fix this by dropping the
original skb owner. However, this is undesirable as the packet has not
physically left the host yet.
In fact, all other tunnels in the kernel do not account skb's passing
through to their own socket. In partciular, ESP over UDP does not do
so and it is the closest tunnel type to PPPoL2TP. So this patch simply
removes the socket accounting in pppol2tp_xmit. The accounting still
applies to control packets of course.
I've also added a reminder that the outgoing checksum here doesn't work.
I suppose existing deployments don't actually enable checksums.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
projects / linux-2.6-omap-h63xx.git / log