Signed-off-by: NJay Fenlason <fenlason@redhat.com>

Prompted by https://bugzilla.redhat.com/show_bug.cgi?id=323411Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

Signed-off-by: NYann Dirson <ydirson@altern.org>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> (update)

fw_node() is not used (and not useful) outside fw-topology.c.
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

This duplicates the read cycle timer feature of raw1394 (added in Linux
2.6.21) in firewire-core's userspace ABI.  The argument to the ioctl is
reordered though to ensure 32/64 bit compatibility.
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: NKristian Høgsberg <krh@redhat.com>

Check NodeID.nodeNumber as per OHCI 1.1 clause 7.2.3.2.  See also IEEE
1394a table 5B-1.

Also, demote the "node ID not valid" message from error to notification
as it is not an error condition.
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

at_context_queue_packet() didn't clean up in an early exit path.
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: NKristian Høgsberg <krh@redhat.com>

Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

Add used includes, remove unused includes.
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

It seems unlikely, but access to self_id_cpu[0] could at least in theory
be deferred until after the loop over self_id_cpu[1..n] or even after
the subsequent reg_read.  Enforce the desired order by a read barrier.

Also prevent the reg_read from being reordered relative to the for loop.
This isn't necessary if the loop's conditional printk counts as an
implicit barrier, but better make it explicit.

(self_id_cpu[] is a coherent DMA buffer.)
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

Firewire-sbp2 did very uncooperative things in the kernel's shared
workqueue:  Sleeping until reception of management status from the
target for up to 2 seconds, and performing SCSI inquiry and all of the
setup of SCSI command set drivers via scsi_add_device.  If there were
transient or permanent error conditions, this caused long blockage of
the kernel's events process, noticeable e.g. by blocked keyboard input.

We now allocate a workqueue process exclusive to fw-sbp2.  As a side
effect, this also increases parallelism of fw-sbp2's login and reconnect
work versus fw-core's device discovery and device update work which is
performed in the shared workqueue.
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: NKristian Høgsberg <krh@redhat.com>

On rare occasions, the ability to set one of the workaround flags at
runtime may save the day.

People who experience I/O errors with firewire-sbp2 while the old sbp2
driver worked for them should try workarounds=1 and report to the devel
mailinglist whether that improves things.  Firewire-sbp2 defaults to the
SCSI stack's maximum transfer size per command, while sbp2 limits them
to 128 kBytes.  Flag 1 accomplishes just that.
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

Fixes "New firewire stack only recognizing half of a chain of drives",
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=242254Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

On IOMMU-less noncoherent architectures, orb->callback will memcpy the
whole SCSI command buffer for READ-like SCSI commands.  It is therefore
friendlier to enable IRQs before the call, like before patch "Add
ref-counting for sbp2 orbs".
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>
Acked-by: NKristian Høgsberg <krh@redhat.com>

Sparse warned about it although it was apparently harmless:

drivers/firewire/fw-cdev.c:624:23: warning: symbol 'interrupt' shadows an earlier one
include/asm/hw_irq.h:29:13: originally declared here
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

Potentially avoids unnecessary loop runs.
Guarantee quadlet-aligned starts of address regions.
Document the return values.
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

drivers/ieee1394/ieee1394_core.c: Define spinlock using
DEFINE_SPINLOCK instead of assignment to SPIN_LOCK_UNLOCKED
Signed-off-by: NMatthias Kaehlcke <matthias.kaehlcke@gmail.com>
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

At least since nodemgr got rid of coarse global locking, accesses to
struct csr1212_keyval's reference counter should be atomic and coupled
with proper barriers.  Also, calls to csr1212_keep_keyval(kv) should
occur before kv is being used.

(We probably should convert refcnt to struct kref, but how to keep
csr1212_destroy_keyval's implementation non-recursively then?)
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

csr1212_keep_keyval(kv) in nodemgr_process_root_directory was
unbalanced if ne->vendor_name_kv already exists.  This happens for
example if eth1394 or raw1394 modify the local config ROM and it is
parsed again.

As a bonus, the attempt to add the vendor_name_kv sysfs attribute
when it already exists is now fixed for good.
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

* Delete optional and empty i2c client_register and client_unregister
  callbacks.
* Use the proper i2c adapter ID.
* Don't use a template to initialize the i2c_adapter structure, it's
  inefficient.
* Update a misleading comment on why we use i2c_transfer rather than
  higher level i2c functions.
Signed-off-by: NJean Delvare <khali@linux-fr.org>
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

The nodemgr host thread can exit on its own even when kthread_should_stop
is not true, on receiving a signal (might never happen in practice, as
it ignores signals). But considering kthread_stop() must not be mixed with
kthreads that can exit on their own, I think changing the code like this
is clearer. This change means the thread can cut its sleep short when
receive a signal but looking at the code around, that sounds okay (and
again, it might never actually recieve a signal in practice).
Signed-off-by: NSatyam Sharma <satyam@infradead.org>
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

sbp2_host_reset and sbp2_handle_status_write are not serialized against
sbp2_alloc_device and sbp2_remove_device.
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

bad_proto can be reached from points which did not take priv->lock.
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

  - Previese patch to NCR5380 broke scsi_mac because
    AUTOSENSE was defined after the inclusion of
    NCR5380.h. Fix it
Signed-off-by: NBoaz Harrosh <bharrosh@panasas.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

There is a test in tgafb_mono_imageblit() for a colour image with a fall-back
to cfb_imageblit().  The test is not necessary as the only caller, which is
tgafb_imageblit(), checks it too and only invokes this function for monochrome
images.  It looks like a left-over from before some changes to
tgafb_imageblit().
Signed-off-by: NMaciej W. Rozycki <macro@linux-mips.org>
Cc: Antonino Daplas <adaplas@pol.net>
Cc: Jay Estabrook <Jay.Estabrook@hp.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The pixel mask calculation in tgafb_mono_imageblit() uses a variable
left-shift on a 32-bit data type by up to 32.  Shifting by the width of a data
type or more produces an unpredictable result according to the C standard.

Rather than widening the data type this fix makes sure the count is between 0
and 31.  The reason is not to penalise 32-bit platforms with operation on a
"long long" type for a marginal case that is meant not to happen (blitting an
image of a zero width).

The reason it has escaped for so long is the Alpha, being purely 64-bit, :-)
does not mask the shift out to 32 bits.  This is a valid implementation --
producing the correct result certainly falls within "unpredictable behaviour".
 It does trigger on MIPS though and it is the recent merge of the TC support
which only enabled the driver for use on anything other than the Alpha.  For
MIPS when the width is 32 the mask ends up being 0 rather than 0xffffffff as
it should be and the frame buffer is not updated.
Signed-off-by: NMaciej W. Rozycki <macro@linux-mips.org>
Cc: Antonino Daplas <adaplas@pol.net>
Cc: Jay Estabrook <Jay.Estabrook@hp.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

As reported in Bugzilla Bug 9093, upon switching to X, a small rectangular
cursor can still be seen blinking in the upper left part of the screen.  It is
fbcon's text cursor.  This is caused by a strange ioctl(..., KDSETMODE,
KD_TEXT) call done by something in userspace, perhaps by X itself, while the
tty is still in graphics mode.  And when the tty is in KD_TEXT mode, the
cursor timer is restarted.

Although this is a userspace problem, we can work around it by delaying the
restart of the cursor timer until an fbcon_switch() is called.  In other
words, the cursor timer will not be restarted even if a KD_TEXT mode switch is
requested.

Regression potential: Present but low
Signed-off-by: NAntonino Daplas <adaplas@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This correct case, when source and destination X coordinates difference is n
multiple of pixels in byte.  This is probably rare case, but this case should
supported for completeness.

Reorganization of FB_READL and FB_WRITEL calls results in code size decrease
for normal build without swapping support and size with support enabled is
reasonable too.

[adaplas]
Add missing fb_rev_pixels_in_long() prototype.
Signed-off-by: NPavel Pisa <pisa@cmp.felk.cvut.cz>
Signed-off-by: NAntonino Daplas <adaplas@gmail.com>
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This patch adds check if selected pixclock is valid (is in the PLL range).

Previously, if the pixclock could not be set, the new mode resolution was set
but pixclock was not set which led to incorrect timings sent to monitor.

[adaplas]
Fixed a few misplaced curly braces.
Signed-off-by: NKrzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: NAntonino Daplas <adaplas@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This patch changes mode selection matching algorithm.  It allows to choose
mode with matching depth even when requested color lengths are greater than
color lengths of every mode with requested color depth.

It also fixes bug in s3fb - wrong error value returned when format is not
supported by chip.
Signed-off-by: NOndrej Zajicek <santiago@crfreenet.org>
Signed-off-by: NAntonino Daplas <adaplas@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Default to the 480p video mode on DVI-D and HDMI displays if `video=safe' is
passed on the kernel command line.

This is intended to be used by `kboot'-style boot loaders (i.e. first-stage
kernels) for the PS3, to provide a failsafe video mode.
Signed-off-by: NGeert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Move the extern declaration for global_mode_option to <linux/fb.h> and rename
the variable to fb_mode_option.
Signed-off-by: NGeert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

ps3fb: Enhance horizontal panning on firmware 1.90 and up:
  - On firmware 1.90 and up, L1GPU_CONTEXT_ATTRIBUTE_FB_BLIT supports copying
    rectangles from XDR to DDR memory with different source and destination
    line lengths. Hence larger horizontal virtual resolutions can be supported
    (up to 16368 pixels).
  - As the actual frame buffer memory layout no longer matches the entries in
    ps3fb_res[], create fake struct ps3fb_ioctl_res data for the
    PS3FB_IOCTL_SCREENINFO ioctl, so user space applications that depend on it
    keep on working.
Signed-off-by: NGeert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

ps3fb: Add virtual screen and panning support:
  - The vertical virtual screen size is limited by the amount of memory
    reserved for ps3fb,
  - The horizontal virtual screen size is limited to the fullscreen width,
  - Advertise that we support panning, so fbcon will use it if the virtual
    screen is enabled.
    Enabling a virtual screen (using `fbset -vyres nnn') can speed up text
    console scrolling by a factor of 10-15, depending on the video mode.
Signed-off-by: NGeert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Don't keep the borders for non-fullscreen modes in XDR memory:
  - Extract ps3fb_sync_image()
  - Work around the alignment restrictions of L1GPU_CONTEXT_ATTRIBUTE_FB_BLIT
    by using an offset with L1GPU_CONTEXT_ATTRIBUTE_DISPLAY_FLIP
  - Only copy the visible part of the screen on every vblank
  - Always put the real frame buffer at the start of video memory
  - Clear fullscreen DDR memory on mode change
Signed-off-by: NGeert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

ps3fb: Use fb_info.par properly:
  o Move mode-specific fields into struct ps3fb_par
  o Allocate struct ps3fb_par using framebuffer_alloc()
  o Protect access to ps3fb_par in ps3fb_sync() using the console semaphore
    (this semaphore is already held when ps3fb_set_par() is called)
  o Avoid calling ps3av_set_video_mode() if the actual video mode hasn't
    changed
Signed-off-by: NGeert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

ps3fb: In the case of non-fullscreen video modes, there was a partial overlap
of the GPU command buffer and the frame buffer. Fix and cleanup various issues
with overlap and alignment:
  - Move the GPU command buffer from the beginning to the end of video memory
  - Exclude the GPU command buffer from the actual frame buffer memory
  - Align the start of the virtual frame buffer to PAGE_SIZE instead of to 64
    KiB, and don't waste memory if it's already aligned (for fullscreen modes)
  - Take into account the alignment when checking memory requirements and
    maximum number of frames
  - Make sure fb_fix_screeninfo.smem_start always points to the virtual frame
    buffer start, so we don't have to compensate for that in ps3fb_mmap()
Signed-off-by: NGeert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Make ps3fb_wait_for_vsync() and ps3fb_flip_ctl() static, as they're no
(longer) used outside ps3fb.
Signed-off-by: NGeert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>