Oleksij Rempel [Thu, 13 Oct 2016 07:41:06 +0000 (09:41 +0200)]
add -Wunused-but-set-variable
we can't just now use -Wall. So lets add them step by step.
-Wunused-but-set-variable seems to be a good start :)
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Adrian Chadd [Tue, 11 Oct 2016 14:24:46 +0000 (07:24 -0700)]
Merge pull request #102 from RyDroid/travis
Adding a file for Travis CI
Adrian Chadd [Sun, 9 Oct 2016 15:54:29 +0000 (08:54 -0700)]
Merge pull request #104 from olerem/pre_debian-2016.10.05
make toolchain path configurable
Paul Fertser [Sat, 1 Oct 2016 16:46:25 +0000 (18:46 +0200)]
make toolchain path configurable
this is needed to make it work with debian
cross-toolchain
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Nicola Spanti (RyDroid) [Tue, 4 Oct 2016 21:56:24 +0000 (23:56 +0200)]
Adding a file for Travis CI
Adrian Chadd [Sun, 2 Oct 2016 18:37:42 +0000 (11:37 -0700)]
Merge pull request #99 from olerem/pre_debian-2016.10.02
Pre debian 2016.10.02
Oleksij Rempel [Sat, 1 Oct 2016 12:10:41 +0000 (14:10 +0200)]
remove unused cmnos_printf.c
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sat, 1 Oct 2016 11:59:13 +0000 (13:59 +0200)]
remove linux/compiler.h, mosty it is not used
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sat, 1 Oct 2016 11:37:54 +0000 (13:37 +0200)]
change licese for atheros files
fix license issues for debian packaging: licensecheck --copyright -r target_firmware
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sat, 1 Oct 2016 11:24:54 +0000 (13:24 +0200)]
change xtansa licesed files
fix license issues for debian packaging: licensecheck --copyright -r target_firmware
With permission from Tensilica, Inc, these files have been relicensed.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sat, 1 Oct 2016 08:03:37 +0000 (10:03 +0200)]
fix missing licenses
fix license issues for debian packaging: licensecheck --copyright -r target_firmware
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sat, 1 Oct 2016 07:33:53 +0000 (09:33 +0200)]
Fix FSF adress in GPLv2 files
fix license issues for debian packaging:
licensecheck --copyright -r target_firmware
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sat, 1 Oct 2016 07:08:57 +0000 (09:08 +0200)]
fix path for a sboot file
sboot was move to other location.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Thu, 29 Sep 2016 19:35:36 +0000 (21:35 +0200)]
use binutils 2.27
to provide debian package we need to use binutils 2.27.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sat, 1 Oct 2016 14:24:32 +0000 (16:24 +0200)]
CMakeLists.txt: add -fno-lto
this option is needed to make compile work with cross-toolchain properly.
We are not using LTO any way, but for some reasons if cross-tolachain
is compailed without LTO, then FW compilation will fail.
With and without this option FW has same md5sum - So it should be save
to use this option.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Adrian Chadd [Tue, 20 Sep 2016 18:17:55 +0000 (11:17 -0700)]
Merge pull request #94 from olerem/crypto
add cryptographic assurances when fetching the toolchain
Daniel Kahn Gillmor [Tue, 20 Sep 2016 17:38:03 +0000 (19:38 +0200)]
add cryptographic assurances when fetching the toolchain
Previously, anyone tampering with the network could have modified the
toolchain source code in transit and the build would have proceeded as
expected.
With this patch, we can ensure that we're getting the expected
tarballs from upstream.
If the required toolchain packages change in the future, the sha256
digests of the new packages should be updated in Makefile as well.
Updated-by: Oleksij Rempel <linux@rempel-privat.de>
Adrian Chadd [Tue, 20 Sep 2016 16:19:48 +0000 (09:19 -0700)]
Merge pull request #93 from olerem/gcc-clean-2016.09.19.1
Gcc clean 2016.09.19.1
Oleksij Rempel [Fri, 16 Sep 2016 08:23:13 +0000 (10:23 +0200)]
fix compile warnings for gcc 6.2.0
Most of issues discovered by gcc 6.2.0 are:
warning: implicit declaration of function '...'
which make sense since this functions was never declared,
espesially functions provided by ROM.
In the last case, correct usage of this functions was never
checked by compiler.
So, wellcome in to GCC 6.2.0 age ;)
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Fri, 16 Sep 2016 06:14:08 +0000 (08:14 +0200)]
Revert "Add a patch for binutils 2.23.1"
This reverts commit
24b475841870cc495417619637c42d65381be38e.
We are using now binutils 2.26.1. So we can drop this patch.
Oleksij Rempel [Thu, 15 Sep 2016 18:11:01 +0000 (20:11 +0200)]
update binutils.patch for 2.26.1
and drop useless update of copy right - it prevent us from
reusing this patch for latest binutils.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Thu, 15 Sep 2016 17:45:09 +0000 (19:45 +0200)]
update gcc.patch for gcc 6.2.0
and drop useless update of copy right - it prevent us from
reusing this patch for latest gcc.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Thu, 15 Sep 2016 18:11:01 +0000 (20:11 +0200)]
bump gcc to 6.2.0
gcc 6.2.0 builds with texinfo versions 6.x .
update other dependencies too.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Adrian Chadd [Fri, 16 Sep 2016 16:10:43 +0000 (09:10 -0700)]
Merge pull request #92 from olerem/date-2016.09.16
remove __DATE__ and __TIME__
Oleksij Rempel [Fri, 16 Sep 2016 07:17:23 +0000 (09:17 +0200)]
remove __DATE__ and __TIME__, make builds reproducible.
__DATE__ and __TIME__ are mostly useless information.
Even more, it makes builds not reprodusable which is evil.
Read more about reproducible builds here:
https://reproducible-builds.org/
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Adrian Chadd [Thu, 11 Aug 2016 20:32:34 +0000 (13:32 -0700)]
Merge pull request #79 from kpengboy/master
Fix license missed during initial conversion
Kevin Peng [Tue, 1 Sep 2015 23:14:31 +0000 (16:14 -0700)]
Fix license missed during initial conversion
Requested by Adrian Chadd
Adrian Chadd [Sun, 16 Nov 2014 02:36:43 +0000 (18:36 -0800)]
Merge pull request #69 from chunyeow/master
ath9k_htc_firmware: check only the mesh control present subfield
Adrian Chadd [Sun, 16 Nov 2014 02:36:17 +0000 (18:36 -0800)]
Merge pull request #67 from IanR778/timestamp_fix
Timestamp fix - looks good! Thanks!
Adrian Chadd [Sun, 16 Nov 2014 02:34:45 +0000 (18:34 -0800)]
Fix the licence - i missed this with my initial conversion
of the licences.
Adrian Chadd [Sat, 15 Nov 2014 20:53:47 +0000 (12:53 -0800)]
Update build instructions for FreeBSD.
Chun-Yeow Yeoh [Fri, 14 Nov 2014 04:50:09 +0000 (12:50 +0800)]
ath9k_htc_firmware: check only the mesh control present subfield
This is related to the previous patch "fix the offset of
CCMP header for mesh data frame". Check only the Mesh
Control Present subfield on QoS Control field to determine
whether it is a mesh data frame.
Signed-off-by: Chun-Yeow Yeoh <yeohchunyeow@gmail.com>
IanR778 [Wed, 8 Oct 2014 22:56:44 +0000 (23:56 +0100)]
Fix: bug in extend_tsf near 0x8000000 values
Bugfix + added more test cases.
IanR778 [Wed, 8 Oct 2014 20:00:37 +0000 (21:00 +0100)]
FIX: support rstamps from the future too.
* Extend a 32 bit TSF to nearest 64 bit TSF value.
* When the adapter is a STATION, its local TSF is periodically modified by
* the hardware to match the BSS TSF (as received in beacon packets), and
* rstamp may appear to be from the future or from the past (with reference
* to the current local TSF) because of jitter. This is mostly noticable in
* highly congested channels. The code uses signed modulo arithmetic to
* handle both past/future cases and signed-extension to avoid branches.
* Test cases:
* extend(0x0000001200000004, 0x00000006) == 0x0000001200000006
* extend(0x0000001200000004, 0x00000002) == 0x0000001200000002
* extend(0x0000001200000004, 0xfffffffe) == 0x00000011fffffffe ! tsfhigh--
* extend(0x00000012fffffffc, 0xfffffffe) == 0x00000012fffffffe
* extend(0x00000012fffffffc, 0xfffffffa) == 0x00000012fffffffa
* extend(0x00000012fffffffc, 0x00000002) == 0x0000001300000002 ! tsfhigh++
IanR778 [Wed, 8 Oct 2014 00:14:01 +0000 (01:14 +0100)]
Firmware sometimes returns erroneous timestamps(2)
Fixed incorrect wrapping handling in ath_extend_tsf().
IanR778 [Wed, 8 Oct 2014 00:11:04 +0000 (01:11 +0100)]
Firmware sometimes returns erroneous timestamps
Added proper synchronization while handling lower 32 bit and upper 32 bit parts in ar5416GetTsf64().
Adrian Chadd [Thu, 11 Sep 2014 18:35:37 +0000 (11:35 -0700)]
Merge pull request #64 from iazz/master
Fix resetting of txstatus flags.
Ignacy Gawędzki [Thu, 11 Sep 2014 09:56:36 +0000 (11:56 +0200)]
Fix resetting of txstatus flags.
Adrian Chadd [Mon, 30 Jun 2014 04:12:00 +0000 (21:12 -0700)]
Merge pull request #61 from sudoman/sudoman-fixes
minimal changes to fix the toolchain build process on some newer systems
Andrew Engelbrecht [Mon, 30 Jun 2014 01:04:09 +0000 (21:04 -0400)]
bump gcc to newer revision
gcc 4.7.4 builds with texinfo versions 5.x .
Andrew Engelbrecht [Mon, 30 Jun 2014 00:56:37 +0000 (20:56 -0400)]
fix binutils build with new gcc versions
disabling -Werror while building binutils ignores the following
warnings on systems using -Werror and -Wall by default:
-Wunused-value
oblique [Wed, 8 May 2013 16:21:31 +0000 (19:21 +0300)]
Add a patch for binutils 2.23.1
Without this patch, binutils fails to compile.
This patch will not be needed for binutils 2.24.
Adrian Chadd [Fri, 23 May 2014 17:00:00 +0000 (10:00 -0700)]
Merge pull request #57 from olerem/io_clean-2014.05.23
Io clean 2014.05.23
Oleksij Rempel [Fri, 23 May 2014 16:36:16 +0000 (18:36 +0200)]
remove all HAL_*_READ/WRITE defines
we do not need them any more
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Mon, 28 Apr 2014 18:08:59 +0000 (20:08 +0200)]
dbg_api.c: use new io funcs
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sun, 27 Apr 2014 14:47:47 +0000 (16:47 +0200)]
cmnos_sflash.c: use new io funcs
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sun, 27 Apr 2014 14:09:00 +0000 (16:09 +0200)]
app_start.c: use new io funcs
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sun, 27 Apr 2014 13:29:58 +0000 (15:29 +0200)]
usb_api_main_patch.c use new io funcs
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sun, 27 Apr 2014 13:27:26 +0000 (15:27 +0200)]
usb_api_k2_patch.c: use new io functions
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sun, 27 Apr 2014 09:25:41 +0000 (11:25 +0200)]
usb_api_magpie_patch.c: use new io funcs
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sun, 27 Apr 2014 14:02:11 +0000 (16:02 +0200)]
init.c: use new io funcs
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sun, 27 Apr 2014 10:13:09 +0000 (12:13 +0200)]
adf_os_io.h: add some ios
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sun, 27 Apr 2014 08:49:35 +0000 (10:49 +0200)]
remove ath_hal_reg_write_target and OS_REG_WRITE
instead use iowrite32_mac
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sun, 27 Apr 2014 08:21:43 +0000 (10:21 +0200)]
remove some unused io defines
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sat, 26 Apr 2014 19:55:14 +0000 (21:55 +0200)]
remove ath_hal_reg_read_target and OS_REG_READ
and use ioread32
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Mon, 28 Apr 2014 14:59:37 +0000 (16:59 +0200)]
init.c: mark wlan_task as noreturn
and remove break state from the loop. It will never jump to flash.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Mon, 28 Apr 2014 10:08:31 +0000 (12:08 +0200)]
make sure app_start is always at the entry point
Add ".boot" section to the linker and make sure app_start()
is always there. Till now every change in app start was cousing
randome issues.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Adrian Chadd [Fri, 23 May 2014 16:01:03 +0000 (09:01 -0700)]
Merge pull request #55 from biergaizi/master
Fix Array Out of Bounds in rcSibUpdate_ht()
Adrian Chadd [Fri, 23 May 2014 15:59:23 +0000 (08:59 -0700)]
Merge pull request #52 from olerem/move_build_2
Move build
Tom Li [Fri, 23 May 2014 15:48:39 +0000 (23:48 +0800)]
Fix Array Out of Bounds in rcSibUpdate_ht()
In rcSibUpdate_ht(), we clear
validPhyRateIndex[WLAN_RC_PHY_MAX][MAX_TX_RATE_PHY]
by a for loop.
But, validPhyRateIndex was defined with
validPhyRateIndex[WLAN_RC_PHY_MAX][MAX_TX_RATE_TBL],
and MAX_TX_RATE_TBL is always greater than MAX_TX_RATE_PHY,
caused the out of bounds array access.
This commit corrects MAX_TX_RATE_PHY to MAX_TX_RATE_TBL,
and remove MAX_TX_RATE_PHY from ratectrl.h because it is
no longer be used.
Signed-off-by: Tom Li <biergaizi@member.fsf.org>
Oleksij Rempel [Mon, 28 Apr 2014 16:36:33 +0000 (18:36 +0200)]
remove some identical usb related headers
k2 and magpie have lots of identical header.
remove some of them.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Mon, 28 Apr 2014 16:22:40 +0000 (18:22 +0200)]
move sboot to the root directory
we do not need it for build any more. But we may need it for research.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Mon, 28 Apr 2014 15:49:40 +0000 (17:49 +0200)]
CMakeLists.txt: do not use include from build
we will need it to avoid confusions
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Mon, 28 Apr 2014 15:48:18 +0000 (17:48 +0200)]
make a local copy of includes needed by magpie
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Adrian Chadd [Fri, 4 Apr 2014 10:38:45 +0000 (03:38 -0700)]
Merge pull request #51 from chunyeow/master
ath9k_htc_firmware: fix the offset of CCMP header for mesh data frame
Adrian Chadd [Fri, 4 Apr 2014 10:36:24 +0000 (03:36 -0700)]
Merge pull request #50 from olerem/testing
testing: collection of different changes
Chun-Yeow Yeoh [Fri, 4 Apr 2014 09:01:15 +0000 (17:01 +0800)]
ath9k_htc_firmware: fix the offset of CCMP header for mesh data frame
Mesh data frame uses at least 3 addresses scheme. So set the offset
correctly while fetching the CCMP header. Previously, we are not
able to do HW encryption for unicast data frame. Applying this
patch will allow HW encryption and avoid loading ath9k_htc.ko using
"nohwcrypt=1".
Signed-off-by: Chun-Yeow Yeoh <yeohchunyeow@gmail.com>
Oleksij Rempel [Sun, 23 Mar 2014 06:48:07 +0000 (07:48 +0100)]
Merge branch 'master' of https://github.com/sdfsdfsdfdsdf/open-ath9k-htc-firmware into testing
Oleksij Rempel [Sun, 23 Mar 2014 06:40:12 +0000 (07:40 +0100)]
Merge branch 'installCmake' of https://github.com/mattflaschen/open-ath9k-htc-firmware into testing
Oleksij Rempel [Sun, 23 Mar 2014 06:34:40 +0000 (07:34 +0100)]
Merge branch 'hif' into next
Oleksij Rempel [Sun, 23 Mar 2014 06:30:53 +0000 (07:30 +0100)]
Merge branch 'speed' into next
sdfsdfsdfdsdf [Sat, 15 Mar 2014 20:32:07 +0000 (20:32 +0000)]
Download through HTTPS for authenticity and privacy (fixes #48)
Oleksij Rempel [Sat, 14 Dec 2013 21:02:48 +0000 (22:02 +0100)]
init.c: do cold reboot on exception
Currently, if we get exception, FW will print exception info, send crash
pattern to the host and reboot. We can reduce deadtime by doing cold reboot
instead of simple reboot. The difference is, that usb interface will be
reseted and usb subsystem will reinit driver. In this case we even do not need
to update host driver.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Matthew Flaschen [Mon, 21 Oct 2013 04:58:20 +0000 (00:58 -0400)]
Note in README that cmake must be installed
Oleksij Rempel [Sat, 14 Sep 2013 15:22:23 +0000 (17:22 +0200)]
split ath_hal_reg_read_tgt and reuse it in ath_hal_reg_rmw_tgt
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sat, 14 Sep 2013 14:44:15 +0000 (16:44 +0200)]
split ath_hal_reg_write_tgt and reuse it in ath_hal_reg_rmw_tgt
make sure we do not have redundand code in ath_hal_reg_rmw_tgt
and ath_hal_reg_write_tgt.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sat, 14 Sep 2013 07:06:38 +0000 (09:06 +0200)]
wmi: add new command WMI_REG_RMW_CMDID
rmw is a compund of reg_read and reg_write. By doint it on FW side
we reduce usb traffic.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Mon, 30 Sep 2013 19:52:12 +0000 (21:52 +0200)]
use only one usb_defs.h
usb_defs.h for k2 and magpie are identical. So it
is better to use one.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Mon, 30 Sep 2013 18:58:28 +0000 (20:58 +0200)]
add new ZM_SOC_USB_DMA_RESET_OFFSET register
add and use ZM_SOC_USB_DMA_RESET_OFFSET
instead if magic number 0x118
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Mon, 30 Sep 2013 14:54:14 +0000 (16:54 +0200)]
usb_api_k2_patch.c: use same style like usb_api_magpie_patch.c
Just other comment styles and tabs instead of spaces.
Some commented A_PRINFs was removed.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Mon, 30 Sep 2013 14:10:33 +0000 (16:10 +0200)]
hif/usb_api: unify platform specific patch file names
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Mon, 30 Sep 2013 10:05:31 +0000 (12:05 +0200)]
hif/usb_api: remove dup code - _fw_usbfifo_recv_command
and _fw_usbfifo_init
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sat, 28 Sep 2013 21:45:48 +0000 (23:45 +0200)]
hif/usb_api: remove dup code - bGet_descriptor_patch
In this patch i also remove eeprom check. It was never
executed any way.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sat, 28 Sep 2013 19:46:52 +0000 (21:46 +0200)]
hif/usb_api: remove dup code - ep6 fix
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Fri, 27 Sep 2013 22:17:41 +0000 (00:17 +0200)]
hif/usb_api: remove dup code - usb_reg_in_patch
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Fri, 27 Sep 2013 21:50:17 +0000 (23:50 +0200)]
hif/usb_api: remove dup code - usb_reg_out_patch
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Adrian Chadd [Fri, 23 Aug 2013 07:48:04 +0000 (00:48 -0700)]
Merge pull request #38 from olerem/next
Fixes for reboot issues
Oleksij Rempel [Tue, 13 Aug 2013 13:53:46 +0000 (15:53 +0200)]
remove transfer type patches for EP3 and EP4
here is comment from kernel source which has enought explanation:
* Some buggy high speed devices have bulk endpoints using
* maxpacket sizes other than 512. High speed HCDs may not
* be able to handle that particular bug, so let's warn...
in our case EP3 and EP4 have maxpacket sizes = 64!!!
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Tue, 13 Aug 2013 10:23:28 +0000 (12:23 +0200)]
patch bcdDevice value for magpie to make easy firmware detection.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Tue, 13 Aug 2013 10:12:15 +0000 (12:12 +0200)]
fix u8 to u16 conversation
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Tue, 13 Aug 2013 09:31:04 +0000 (11:31 +0200)]
patch bcdDevice value for k2 to make easy firmware detection.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Tue, 13 Aug 2013 09:23:14 +0000 (11:23 +0200)]
help linker to find u8UsbDeviceDescriptor on k2
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Wed, 24 Jul 2013 08:26:18 +0000 (10:26 +0200)]
k2_fw_usb_api: workaround for EP4 bug.
This patch is workaround for xhci related issue.
After usb port resume FIFO on EP4 can't accept Bulck stream longer then
64 Bytes.
Currently we have no better way then just completly reset chip.
Side effect of this workaround - disconnection event for this device.
It means, module will be reloaded and we will get new usb minor id.
This is why, we should not do it on each firmware restart.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Fri, 19 Jul 2013 18:37:36 +0000 (20:37 +0200)]
docs: add chip images for documentation
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Thu, 18 Jul 2013 14:44:46 +0000 (16:44 +0200)]
k2_fw_usb_api: prevent buffer overflow.
This was reproduced on intel USB 3.0 controller.
After getting corrupt packet we was jumping bejond allocated buffer.
Insted of oopsing we can at lest warn hier.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Thu, 18 Jul 2013 13:02:52 +0000 (15:02 +0200)]
k2_fw_usb_api: add warning on oversized buffer.
Currently i do not know how to fix this bug. It is reproducable on:
Intel Corporation 7 Series/C210 Series Chipset Family USB xHCI Host Controller (rev 04)
After module reload, first packet with data size over 64 Byte will brake EP4
buffer. Size register will show bigger value as existing size of buffer.
Last buffer part will be corrupt.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Sun, 7 Jul 2013 18:43:07 +0000 (20:43 +0200)]
if_ath.c: make ath_hal_reg_write_tgt more readable
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Fri, 5 Jul 2013 19:33:31 +0000 (21:33 +0200)]
k2_fw_usb_api: fix typo in usb reg address
we was writing to wrond register.
0x100ae = ep3
0x100af = ep4
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Fri, 5 Jul 2013 19:31:43 +0000 (21:31 +0200)]
k2_fw_usb_api: fix typo in reg address
this was writing defenetly to wrong address.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Oleksij Rempel [Wed, 3 Jul 2013 18:45:50 +0000 (20:45 +0200)]
k2_fw_usb_api: add delay after mac reset
We need some time between MAC reset and next operation. This will
avoid some freezes on module and firmware reload.
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
projects / librecmc / open-ath9k-htc-firmware.git / log