Age | Commit message (Collapse) | Author | Files | Lines |
|
the chip can only handle aligned DMA address (4byte align) for
rx-buffers,
skb_reserve(skb, NET_IP_ALIGN) will break the rule, so, just remove it.
Signed-off-by: xiong <xiong@qca.qualcomm.com>
|
|
when reset in alx_task/alx_reinit, alx_halt function calls
cancel_work_sync(task), which makes a dead loop. This patch fixes it by
adding parameter of in_task for alx_halt & alx_reinit.
Signed-off-by: xiong <xiong@qca.qualcomm.com>
|
|
Fixes build breakage in Ubuntu PPC build, e.g.:
alx_main.c: error: implicit declaration of function ‘vzalloc’ [-Werror=implicit-function-declaration]
Signer-off-by: Kamal Mostafa <kamal@canonical.com>
|
|
Use of __dev{init,exit}* is deprecated.
Signed-off-by: Kamal Mostafa <kamal@canonical.com>
|
|
the clock for MDIO should be slow if link down, and be fast if link on.
Signed-off-by: xiong <xiong@qca.qualcomm.com>
|
|
define new registers value for C0 version for register checking test,
and untouch PHY setting when do MAC loopback test.
PS. PHY external loopback isn't supported for FPGA due to board
limitation.
Signed-off-by: xiong <xiong@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
because FULL_DUPLEX & DUPLEX_FULL are different, 'ethtool ethX' will show
wrong duplex info, this patch fix it.
the PHY initial advertisement is also changed to reflect real HW capability.
Signed-off-by: xiong <xiong@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
256 table size will casue wrong CPU/Q number in RRD description.
Signed-off-by: xiong <xiong@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
ONLY 1 napi is allocated for MSI & INTx mode interrupt, so,
only 1 txq/rxq is supported.
Signed-off-by: xiong <xiong@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
current C0 FPGA platform is ready, this patch adds new register difinition
for C0.
To run this driver on FPGA, just revised hw->is_fpga to 'true' in init_sw
Signed-off-by: xiong <xiong@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
loopback (both MAC and PHY) test is added for self-test.
Signed-off-by: xiong <xiong@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
MSIX, MSI, and INTx interrupt modes are tested, if the platform support.
Signed-off-by: xiong <xiong@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
MTU ought to be set whether the interface is 'Up' or 'Down'.
Signed-off-by: xiong <xiong@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
basic function of hw-stats and self-test is added.
Signed-off-by: xiong <xiong@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
tx-timeout after sleep/resume is caused by wrong interrupt mode used
after resume.
Signed-off-by: xiong <xiong@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
version format : MAJ.MIN.PATCH
Signed-off-by: xiong <xiong@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
WoL speed is wrongly set to 100M, revise it to 10M as 1st priority for
powersaving.
Signed-off-by: xiong <xiong@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
Xiong is sending out other patches that split this one
into other smaller patches.
This reverts commit fcbfd447c62159501fb817eac6cd99ea8577975f.
|
|
this patch fixed followking issues:
1. WoL seed is wrongly set to 100M, revise it to 10M.
2. tx-timeout after sleep/resume. it's caused by wrong interrupt mode.
3. add stats/diag features for ethtool.
TODO:
the diag(self-test) function is very simple for this version,
we will add loopback/interrupt test for next version.
Signed-off-by: xiong <xiong@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
this patch include:
1. split register definitions to alx_reg.h.
2. MAC/DMA is reset when PHY link change Down.
3. fixed all error/warnings found by checkpatch.
Signed-off-by: xiong <xiong@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
The LICENSE is ISC and some original code came from me based
on the compat project, I have relicensed that stuff to ISC
and contributed here. The rest is the QCA alx driver which
is all ISC licensed.
The documentation for submitting patches and using the
Signed-off-by is added. This follows the Linux kernel's
Developer's Certificate of Origin 1.1. We don't yet have
a mailing list but what we do have is commitment to support
this by the QCA alx development team. Adrian has offered
to help address the BSD target support.
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
Provide clear instructions as to what is required.
To make things easier for users make the linux target
depend on the linux-src target.
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
While at it update the Linux help targets.
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
This is an implicit target not exposed to the user
through the help menu but used within the port Makefile
to separate backporting from a clean linux-next target
build.
This also calls ./refresh-compat for us.
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
This should make it cleaner at the top level to
understand what belongs to what OS.
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
compat-drivers had removed alx from being cherry picked
as a crap type patch. The alx driver is not yet upstream
and a compromise has been reached to try to help enable
users with the alx module and to help address killing
proprietary drivers. This unified alx git tree will be
used to supply the target OS support for Linux and FreeBSD,
and compat-drivers will cherry pick the alx driver from
here while it is not yet upstream as the project aims to
provide a foundation for how to do open unified driver
development.
This also adds the -DCOMPAT_BASE which is now required
on the latest compat-drivers.
The changes in this commit address the latest changes
to compat-drivers to support unified drivers.
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
This clarifies this is under the ISC license.
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
1. code are re-orgnized, abstract layer is removed. struct alx_hw is removed.
2. Multi-TX feature is added since the HW have 4 tx-Qs.
3. Interrupt usage for MSIX for RSS/Multi-TX is revised to share vector to
save system resources.
Interrupt allocation is moved to .probe routine.
4. Extension PHY resgister access is added as well.
please NOTE, this version of code just pass compiler (under net-next.git env),
only load/unload test is done. any other network related functions are not
tested yet.
BTW, when I run checkpatch on this code, there are many warnings about:
WARNING: networking block comments put the trailing */ on a separate line
there are also many many same warnings when I run it on other drivers in the
kernel such as broadcom/tg3, intel/igbe
Signed-off-by: xiong <xiong@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|
|
Goal is to use modern tools like coccinelle to see if we can
unify the alx Ethernet driver between Linux and FreeBSD.
The objective is to see if in practice even with these
tools we can keep getting updates to both OSes through
a single tree but at the same time see what challenges
we face.
This is part of the lofty effort on killing proprietary drivers.
I have taken the alx driver code from public postings on public
mailing lists. At this point I have not updated the alx driver
to match the newest v2 patch iterations. Work is still being
done to modify that driver to meet community standards.
Ideally we should try to first complete the driver with FreeBSD
support here prior to getting it upstream in the Linux kernel
as it would make the development easier. Once we have support
targets for both FreeBSD and Linux we can push upstream.
If we don't care then forget about this project.
Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
|