Fix the transmit function for the following:
* Free XmtCmd in the error code path. This use to leak memory in
error conditions.
* Do pci mapping after the checksum operations are over. They can
reallocate the skb at a different location.
* Fix UDP checksum errors which were seen in wireshark
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Leave a spinlock before calling request_irq(). request_irq() calls kmalloc
which can sleep. This was generating a warning dump while driver is loaded.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Use correct queue_id while transmitting non-TCP packets.
They should always use queue 0.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
The watchdog timer which updates the link status was not fired at the
end of sxg_entry_open(). Add that.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Alexander Beregalov <a.beregalov@gmail.com>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Replace all DMA_32BIT_MASK macro with DMA_BIT_MASK(32)
Signed-off-by: Yang Hongyang<yanghy@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Replace all DMA_64BIT_MASK macro with DMA_BIT_MASK(64)
Signed-off-by: Yang Hongyang<yanghy@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Add support for downloading the firmware using kernel-builtin mechanism.
This will remove the need for the firmware files in the driver source code.
Signed-off-by: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Add a watchdog timer to take care of link change notifications.
Link changes would now be handled asynchronously as they involve large delays.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix a softlock warning message thrown up by the rtnl_timer.
This was observed during the LTP tests (interface up/down test)
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Read the correct MAC address from EEPROM/Flash.
Without this fix, all the interfaces were using the same MAC address.
This works on Rev B firmware.
Signed-off-by: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix the problem of the interface not working after a sequence of up-down-up
events. The problem was observed only on systems where the card was using MSI-X
interrupts. On the second time open, the driver did not request MSI-X vector
but was trying to use normal interrupts.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* Update the module name in Kconfig help section.
This is a classic case of documentation keeping out of pace with
development
And this was overlooked by me ages ago when we had fixed the Makefile for
sxg_ethtool to compile.
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Cc: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* The receive rings were getting allocated twice. Once at probe time
and once at open time. This leaked huge amounts of memory.
Fix this leak and now allocation is done only once.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* This patch adds support for controling checksum feature using the ethtool
interface.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Remove the deprecated microcode which was for Rev A.
Now on the driver will use Rev B microcode only.
Signed-off-by: Michael Miles <mmiles@alacritech.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch makes the Sahara SXG driver use Rev B firmware instead of Rev A.
The firmware version is 1.71
Signed-off-by: Michael Miles <mmiles@alacritech.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch adds the new firmware files required by Rev B cards.
Signed-off-by: Michael Miles <mmiles@alacritech.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* Fix a memory leak if allocation of skb fails in sxg_fill_descriptor_block.
In sxg_fill_descriptor_block(), if allocation of skb failed in loop,
we just came out shouting. This rollbacks all the successful operation
before skb allocation fails and then returns.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* This patch adds support for offloading checksum to hardware.
IP checksum have been tested for IPv4 and IPv6.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix staging/sxg napi interface calls:
drivers/staging/sxg/sxg.c:1271: error: implicit declaration of function 'netif_rx_schedule_prep'
linux-next-20090209/drivers/staging/sxg/sxg.c:1272: error: implicit declaration of function '__netif_rx_schedule'
drivers/staging/sxg/sxg.c:1325: error: implicit declaration of function 'netif_rx_complete'
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Mithlesh Thukral <mithlesh@linsyssoft.com>
Cc: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Cc: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Make Sahara SXG driver use MSI-X interrupts instead of line based interrupts
if possible. In case of problems in getting MSI-X vectors or MSI-X not being
supported, driver will fall back to use previous line based interrupts.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix the locking issue of locks in transmit code path.
There was an unnecessary check for interrupt context in transmit code path.
Removed that.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* Add NAPI support for SXG driver for Alacritech's 10Gbe products.
The driver will now work in NAPI mode by default.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* Add multicast support for SXG driver for Alacritech's 10Gbe products.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Mithlesh Thukral <mithlesh@linsyssoft.com>
Cc: Michael Miles <mmiles@alacritech.com>
Cc: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This compiles out some functions that are not being used to keep the
build clean so that we can see the "real" warnings and errors.
Cc: Mithlesh Thukral <mithlesh@linsyssoft.com>
Cc: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Cc: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
These variables are never used, so #ifdef them away.
This should probably be fixed up properly from someone who knows why we
are even including these files in the first place, when they do not seem
to be needed at all.
Cc: Mithlesh Thukral <mithlesh@linsyssoft.com>
Cc: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Cc: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
They are not needed here and only cause build warnings and bloat the
object file.
Cc: Mithlesh Thukral <mithlesh@linsyssoft.com>
Cc: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Cc: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This fixes some build problems with the ethtool support recently added
for the sxg driver.
This is a very good example of why you need to check the build for
warnings and then fix them...
Cc: Mithlesh Thukral <mithlesh@linsyssoft.com>
Cc: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Cc: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* Fix problem of crash on 50MB machine.
* Fixed dma_addr_t bug, which resolves issues on x86_32 bit machines.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* Add new firmware and remove all firmware file.
* Add a switch to load either debug or free firmware.
Signed-off-by: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Misc. cleanups in the driver.
* Remove debugging code and variables.
* Fix compile time warnings.
* Remove debugging comments.
* Start cleanup of sxg_stats structure. This structure will eventually become
very small
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* Fix locking related issues like taking locks at right level.
* Convert some variables to atomic, to prevent taking them while
incrementing or decrementing them.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* Cleanup in allocation of SXG_SGLs.
* Locking issues related to SglQLock.
* XmtCmd and XmtZeroLock consistency fixes.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* Add Ethtool framework to driver
* Makefile changes to fix build redundancy.
* Fix ups to error code paths in receieve buffer allocation as well as receive code path.
* Read MAC address from FLASH/EEPROM
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* Cleanup in recevive buffer structure
* Drop receive data buffer as its not needed with use of skbs
* Fix error code paths in receive skb failures
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix up the indentation to Linux style. There was some indentation which was
not as per Linux style specially related to 80 char lines.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch cleans up the comment. Converts the comments to C89 style.
Fixes comment related TODO item.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch removes all typedefs in the code. These were the typedefs
which are still present in driver in staging tree after the cleanup patches.
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* This patch introduces the new SXG_SGL design.
* Related changes to sxg_scatter_gather structure.
* Introduced PSXG_X64_SGL changes which are x64 friendly
* Setting the MAC HEADER pointer properly in skb before giving to higher
layers.
Signed-off-by: Michael Miles <mmiles@alacritech.com>
Signed-off-by: LinSysSoft Sahara Team <saharaproj@linsyssoft.com>
Signed-off-by: Christopher Harrer <charrer@alacritech.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Remove ugly 'Stop Comilation;' statement in sxghif, replace with an
error macro.
This should never be hit as we are only building for x86 boxes at the
moment.
Signed-off by: J.R. Mauro <jrm8005@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
These functions aren't used yet, so put them behind the
proper #define so the compiler doesn't complain about them.
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Change C99 comments to C89 comments
Some nested comments seem to have been missed and some blocks are redundantly
commented, but at least most of the //'s are gone
Signed-off by: J.R. Mauro <jrm8005@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
__FUNCTION__ is gcc-specific, use __func__
Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Mithlesh Thukral