While porting GRO to r8169, I found this driver has a bug in its rx
path.
All skbs given to network stack had their ip_summed set to
CHECKSUM_NONE, while hardware said they had correct TCP/UDP checksums.
The reason is driver sets skb->ip_summed on the original skb before the
copy eventually done by copybreak. The fresh skb gets the ip_summed =
CHECKSUM_NONE value, forcing network stack to recompute checksum, and
preventing my GRO patch to work.
Fix is to make the ip_summed setting after skb copy.
Note : rx_copybreak current value is 16383, so all frames are copied...
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix fb_defio bug in udlfb for multiple framebuffers
Previously when using fb_defio with multiple DisplayLink attached displays
udlfb incorrectly used a shared datastructure where it needed to have
one per fb_info struct. This was completely broken and caused all kinds of
problems, including rendering to one framebuffer affecting others.
Signed-off-by: Bernie Thompson <bernie@plugable.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The Kconfig description for CONFIG_SBE_PMCC4_NCOMM says:
If you have purchased this optional support you must say Y or M
here to allow the driver to operate with the NCOMM product.
This is not correct as it can not be built as module,
this patch updates the description accordingly.
Signed-off-by: Michael Prokop <mika@grml.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
In each case, the function has an unsigned return type, but returns a
negative constant to indicate an error condition. For move_block_from_dma,
there is only one call and the return value is dropped, so it need not be
unsigned. For labpc_eeprom_write, there is only one call and the result is
stored in a signed variable, so again the unsigned return type is not
necessary.
A sematic match that finds this problem is as follows:
(http://coccinelle.lip6.fr/)
// <smpl>
@exists@
identifier f;
constant C;
@@
unsigned f(...)
{ <+...
* return -C;
...+> }
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
kmalloc() may fail, check for it.
Allocated memory is not freed.
Use IS_ERR() instead of strict checking.
Signed-off-by: Vasiliy Kulikov <segooon@gmail.com>
Cc: David Cross <david.cross@cypress.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Functions alloc_ieee80211 and register_netdev may fail. Check for it.
Signed-off-by: Vasiliy Kulikov <segooon@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Add module options for console and fb_defio
Convert fb_defio on/off switch to module option and add console option.
>From the command line, pass options to modprobe
modprobe udlfb defio=1 console=1
Or for permanent option, create file like /etc/modprobe.d/options with text
options udlfb defio=1 console=1
Accepted options:
fb_defio Make use of the fb_defio (CONFIG_FB_DEFERRED_IO) kernel
module to track changed areas of the framebuffer by page faults.
Standard fbdev applications that use mmap but that do not
report damage, may be able to work with this enabled.
Disabled by default because of overhead and other issues.
console Allow fbcon to attach to udlfb provided framebuffers. This
is disabled by default because fbcon will aggressively consume
the first framebuffer it finds, which isn't usually what the
user wants in the case of USB displays.
Signed-off-by: Bernie Thompson <bernie@plugable.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Support writing default/backup edid via /sys/class/graphics/fb?/edid
Writing a backup edid blob to the 'edid' file will cause udlfb to
attempt to re-read the edid from hardware.
If we still don't have a good edid, it will use the backup edid instead.
The intent is to 1) enable user-supplied edid
2) enable udev rules to backup the last known good edid
for a device, and then provide that edid as a backup for all future starts
Signed-off-by: Bernie Thompson <bernie@plugable.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix big endian rendering bug (affects PowerPC)
Thanks to Christian Melki at Ericsson for finding and suggesting patch.
Signed-off-by: Bernie Thompson <bernie@plugable.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Enhance EDID and mode table handling
* Adds support for lower-end DL chips on higher-end monitors
* Adds support for using fbdev standard EDID parsing and handling
* Adds support for more dynamic reallocation of framebuffer, based on new EDID
Queries EDID from hardware with several retries then
queries DisplayLink chip for max pixel area via vendor descriptor.
Uses existing fbdev mode handling service funcations to build mode list
of common modes between adapter and monitor.
This enables lower-end chips (e.g. DL-125) to successfully pick a compatible
mode on higher end monitors (e.g. whose default mode is over 1280x1024).
Reallocate framebuffer if necessary in response to new max mode.
Signed-off-by: Bernie Thompson <bernie@plugable.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Revamp reference handling and synchronization for unload/shutdown
Udlfb is a "virtual" framebuffer device that really exists on
two separate stacks: at the bottom of the framebuffer interface,
and on top of USB. During unload, there's no guarantee which
one will tear down first. So reference counting must be solid
to handle all possibilities and not access anything once its gone.
Signed-off-by: Bernie Thompson <bernie@plugable.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Remove metrics_misc sysfs attribute
Previously returned multiple values, which goes against sysfs guidelines.
Had been used for debugging to figure out which of the multiple
fbdev interfaces an app was using, but log file messages are sufficient.
Signed-off-by: Bernie Thompson <bernie@plugable.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Add ability to power off monitor with standard blanking interface
Used by X servers and others to manage power of dislpay
Signed-off-by: Bernie Thompson <bernie@plugable.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Restore fbdev character read and write support
Enables access to the framebuffer from simple reads and writes
to /dev/fb0 (or fb1, ...). Low performance, because of extra copy
and because udlfb must scan every pixel for changes. But very
useful for testing and simple image display, etc.
Signed-off-by: Bernie Thompson <bernie@plugable.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The Bamboo Touch reports a sub-screen resolution of 480x320. The
signal-to-noise ratio is only about 100, so filtering is needed in
order to reduce the jitter to a usable level. However, the low
resolution leads to round-off errors in the EWMA filter, resulting in
extremely jerky pointer motion. This patch explicitly sets a higher
resolution for those devices, and tells this to the completion handler
via a low-resolution quirk.
Signed-off-by: Henrik Rydberg <rydberg@euromail.se>
Acked-by: Ping Cheng <pingc@wacom.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Add support for the Bamboo Touch trackpad, and make it work with
both the Synaptics X Driver and the Multitouch X Driver. The device
uses MT slots internally, so the choice of protocol is a given.
Signed-off-by: Henrik Rydberg <rydberg@euromail.se>
Acked-by: Ping Cheng <pingc@wacom.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Collect device-specific code into a single function, and use quirks to
flag specific behavior instead.
Signed-off-by: Henrik Rydberg <rydberg@euromail.se>
Acked-by: Ping Cheng <pingc@wacom.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
The Bamboo devices have multiple interfaces which need to be setup
separately. Use the HID parsing mechanism to achieve that.
Signed-off-by: Ping Cheng <pinglinux@gmail.com>
Signed-off-by: Henrik Rydberg <rydberg@euromail.se>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
The signal-to-noise ratio varies between devices, but currently all
devices are treated the same way. Add fuzz parameters to the feature
struct, allowing for tailored treatment of devices.
Signed-off-by: Henrik Rydberg <rydberg@euromail.se>
Acked-by: Ping Cheng <pingc@wacom.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Add the PowerOn (PonKey) button support to detect power on/off events.
Acked-by: Linus Walleij <linus.walleij@stericsson.com>
Signed-off-by: Sundar R Iyer <sundar.iyer@stericsson.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Add logic to wm97xx_read_aux_adc() to retry reading the adc if the
sample failed. This could occur if the previous sample was still in
the return register or the sample timed-out. Also avoid a pathologic
failure mode by disabling the digitizer and returning -EBUSY after 5
retries.
Signed-off-by: Eric Millbrandt <emillbrandt@dekaresearch.com>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Use platform_device_del() instead of platform_device_unregister() in error
handling path.
Signed-off-by: Axel Lin <axel.lin@gmail.com>
Acked-by: Liam Girdwood <lrg@slimlogic.co.uk>
Acked-by: Mark Brown <broonie@opensoruce.wolfsonmicro.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Return recovered from pci_mmio_enabled function if it is 82XX.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
The sg table size is increased from 128 to 1024 to support multiple sg lists
and number of sg elements per scsi command.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
The clearing of local references to rport was blocked when abort is
active. This code clears the local references unconditionaly upon
device loss timeout callback from FC transport.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
In case the MPI times out, the FW issues an async event AE8002
to indicate this to every FCoE function. The FC/FCoE driver is
required to handle this, by doing a soft reset and issuing a
Write MPI register mailbox command to reset the MPI.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Currently the IOs are returned back in slot reset, this
could be late and can cause error handler to invoke.
If error handler invokes, eh_abort fails and escalate to
device/bus/host resets causing issues.
The commands need to be returned back to upper layer in
io_error_detected.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
For ISP82xx, the check for empty slot in request queue before posting command type 6
request was missing. This could lead to request queue entry corruptions causing
IO timeouts.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Currently, if target sets the SCSI Status (with Check condition)
and there is no FCP residual bit set then driver does not check
for dropped frame. This could lead to data corruption.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
The fw_hung flag should be set ir-respective of if there is a
mbx command pending or not. Also the complete should be called
if there is a mbx waiting.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
The seconds_since_last_heartbeat should be checked for consecutive
heartbeat checks. Currently it could happen that seconds_since_last_heartbeat
gets set to max (2 seconds) for non-consecutive heartbeat checks.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
We have been seeing the flush request timeout with a wide
range of hardware from tgt+iser to FC targets from a major vendor.
After discussions about if the value should be configurable and
what the best value should be, this patch just increases the flush/sync
cache timeout to 1 minute. 2 minutes was determined to be too long, and
making it configurable was troublesome for users.
This patch was made over Linus's tree. It is not made over scsi-misc
or scsi-rc-fixes, because Linus's had block layer changes that my
patch was built over.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Acked-by: Jens Axboe <jaxboe@fusionio.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
In this code, 0 is returned on memory allocation failure, even though other
failures return -ENOMEM or other similar values.
A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)
// <smpl>
@@
expression ret;
expression x,e1,e2,e3;
@@
ret = 0
... when != ret = e1
*x = \(kmalloc\|kcalloc\|kzalloc\)(...)
... when != ret = e2
if (x == NULL) { ... when != ret = e3
return ret;
}
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
Acked-by: "Desai, Kashyap" <Kashyap.Desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>