The rtl8169_intr_mask variable isn't used anywhere in the code, so
just remove it to avoid a GCC 6.2 compiler warning.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
The call to flush cache on the transmit buffer was misplaced (for very
short packets) and asked to flush less than a cacheline.
Move the flush cache call to after a short packet has been padded
to minimum length (so the padding is flushed too), and round the size
up to a cacheline.
Signed-off-by: Peter Chubb <peter.chubb@data61.csiro.au>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Return value of rtl_send_common propogates unmodified all the way
up to eth_send and further to API consumer if CONFIG_API is enabled.
Previously rtl_send_common returned number of bytes sent on success
which was erroneouly detected as error condition by API consumers
that checked for operation success by comparing return value with 0.
Switch rtl_send_common to use common convention: return 0 on success
and negative value for failure.
Cc: Stephen Warren <swarren@nvidia.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Signed-off-by: Oleksandr Tymoshenko <gonzo@bluezbox.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
The rtl8169 driver uses a global variable to store the register address
of the adapter being operated upon. This is updated to point at the
correct adapter when sending or receiving a packet, or shutting down the
adapter, but not when initializing the adapter. Consequently, switching
between different adapters within the same U-Boot runtime does not work
correctly since the driver programs the wrong registers during
rtl8169_eth_start() -> rtl8169_common_start() -> rtl8169_hw_start().
Note that since rtl8169_eth_stop() does set the global variable, the
second consecutive attempt to use the "new" adapter did work even before
this patch, because each time network usage is shut down, the network
core calls stop, which sets the variable so that the next start does
actually initialize the hardware, and the adapter works.
Equally, rtl8169_eth_probe() calls rtl_init() which sets the global, so
if using only a single device, or if picking the "right" device (based on
probe order) when multiple devices are present, ioaddr will already be set
correctly from the get-go, so the issue does not occur.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
When DEBUG_RTL8169 is on, a build error occurs in function
'rtl_init': error: 'dev' undeclared. Fix this.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Update this driver to use the proper driver-model PCI API functions.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Most driver model PCI functions have a dm_ prefix. At some point, when the
old code is converted to driver model and the old functions are removed, we
will drop that prefix.
For consistency, we should use the dm_ prefix for all driver model
functions. Update pci_get_bdf() accordingly.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Casting from dev->priv to pci_dev_t changes the value's size on a 64-bit
system. This causes the compiler to complain about casting a pointer to an
integer of a different (smaller) size. To avoid this, cast to an integer
of matching size first, then perform an int->int cast to perform the size
change. This signals explicitly that we do want to change the size, and
avoids the compiler warning. This is legitimate since we know the pointer
actually stores a small integer, not a pointer value.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Now that we have a new header file for cache-aligned allocation, we should
move the stack-based allocation macro there also.
Signed-off-by: Simon Glass <sjg@chromium.org>
This driver is used by the Intel Minnowmax board. Convert it to driver model
so it can use the new Ethernet implementation.
Signed-off-by: Simon Glass <sjg@chromium.org>
Turn ioaddr into an unsigned long rather than a sized 32-bit variable.
While at it, fix a couple of pointer to integer cast size mismatch
warnings by casting through unsigned long going from pointers to
integers and vice versa.
Cc: Joe Hershberger <joe.hershberger@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Update the naming convention used in the network stack functions and
variables that Ethernet drivers use to interact with it.
This cleans up the temporary hacks that were added to this interface
along with the DM support.
This patch has a few remaining checkpatch.pl failures that would be out
of the scope of this patch to fix (drivers that are in gross violation
of checkpatch.pl).
Signed-off-by: Joe Hershberger <joe.hershberger@ni.com>
Acked-by: Simon Glass <sjg@chromium.org>
This network interface card is found on the NVIDIA Jetson TK1.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
To work around potential issues with explicit cache maintenance of the
RX and TX descriptor rings, allocate them from a pool of uncached memory
if the architecture supports it.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
RX and TX descriptor rings should be aligned to 256 byte boundaries. Use
the DEFINE_ALIGN_BUFFER() macro to define the buffers so that they don't
have to be manually aligned later on. Also make sure that the buffers do
align to cache-line boundaries in case the cache-line is higher than the
256 byte alignment requirements of the NIC.
Also add a warning if the cache-line size is larger than the descriptor
size, because the driver may discard changes to descriptors made by the
hardware when requeuing RX buffers.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
According to the top-level README file, this configuration setting can
be used to override the number of receive buffers that an ethernet NIC
uses.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
When network protocol errors occur (such as a file not being found on a
TFTP server), the processing done by the NetReceive() function will end
up calling the driver's .halt() implementation. However, after that the
device no longer has access to the memory buffers and will cause errors
such as this in the rtl_recv() function when trying to hand descriptors
back to the device:
pci_hose_bus_to_phys: invalid physical address
This can be fixed by deferring processing of network packets until the
descriptors have been handed back. That way rtl_halt() tearing down
network buffers is not going to prevent access to the buffers.
Reported-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
This chip is compatible with the existing driver, except that it uses
BAR2 instead of BAR1 for the I/O memory region. Using this patch I can
use the PCIe ethernet interface on the CompuLab Trimslice to boot from
the network.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Patch: 276477
Instead of directly calling the low-level invalidate_dcache_range() and
flush_cache() functions, provide thin wrappers that take into account
alignment requirements.
While at it, fix a case where the cache was flushed but should have been
invalidated, two cases where the buffer data was flushed instead of the
descriptor and a missing cache invalidation before reading the packet
data that the NIC just wrote to memory.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Patch: 276474
This chip is compatible with other RTL8168 chips and can be found on the
NVIDIA Cardhu and Beaver boards.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Patch: 276475
currticks() is defined as get_timer(0), which returns an unsigned long,
so use %lu instead of %d to print the result.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Patch: 276473
Fix this:
rtl8169.c: In function 'rtl8169_initialize':
rtl8169.c:907:13: warning: assignment from incompatible pointer type
Signed-off-by: Joe Hershberger <joe.hershberger@ni.com>
Fix:
rtl8169.c: In function 'rtl_init':
rtl8169.c:742:13: warning: variable 'printed_version' set but not used
[-Wunused-but-set-variable]
Signed-off-by: Wolfgang Denk <wd@denx.de>
rtl8169 does not have write_hwaddr function.
However, eth stuff executes write_hwaddr function
because eth_device structure has not been initialized.
Signed-off-by: Nobuhiro Iwamatsu <iwamatsu@nigauri.org>
CC: Ben Warren <biggerbadderben@gmail.com>
When PCI device use system memory, some PCI host controller should be
set physical memory address.
Signed-off-by: Yoshihiro Shimoda <shimoda.yoshihiro@renesas.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Fix the problem that cannot access actual data when CPU data cache enabled.
Signed-off-by: Yoshihiro Shimoda <shimoda.yoshihiro@renesas.com>
Tested-by: Nobuhiro Iwamatsu <iwamatsu.nobuhiro@renesas.com>
Acked-by: Ben Warren <biggerbadderben@gmail.com>
The environment is the canonical storage location of the mac address, so
we're killing off the global data location and moving everything to
querying the env directly.
The drivers that get converted here:
3c589
4xx_enet
dc2114x
dm9000x
enc28j60
fsl_mcdmafec
ks8695eth
mcffec
rtl8019
rtl8169
s3c4510b_eth
xilinx_emac
xilinx_emaclite
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
CC: Ben Warren <biggerbadderben@gmail.com>
CC: Rolf Offermanns <rof@sysgo.de>
CC: Stefan Roese <sr@denx.de>
CC: Sascha Hauer <saschahauer@web.de>
CC: TsiChung Liew <Tsi-Chung.Liew@freescale.com>
CC: Greg Ungerer <greg.ungerer@opengear.com>
CC: Xue Ligong <lgxue@hotmail.com>
CC: Masami Komiya <mkomiya@sonare.it>
CC: Curt Brune <curt@cucy.com>
CC: Michal SIMEK <monstr@monstr.eu>
This patch fixes a problem that RTL8110SCL started transfer
with an incorrect memory address.
Signed-off-by: Yoshihiro Shimoda <shimoda.yoshihiro@renesas.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
When rtl_recv() of rtl8169 is called, OWNbit of status register
is not enable occasionally.
rtl_recv() doesn't work normally when the driver doesn't do
appropriate processing.
This patch fix this problem.
Signed-off-by: Nobuhiro Iwamatsu <iwamatsu@nigauri.org>
Change return values of init() functions in all Ethernet drivers to conform
to the following:
>=0: Success
<0: Failure
All drivers going forward should return 0 on success. Current drivers that
return 1 on success were left as-is to minimize changes.
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Acked-by: Stefan Roese <sr@denx.de>
Acked-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
Acked-by: Kim Phillips <kim.phillips@freescale.com>
Acked-by: Haavard Skinnemoen <hskinnemoen@atmel.com>
Acked-By: Timur Tabi <timur@freescale.com>
It is unclear on what platforms this driver has been tested, since
noone up to now defines CONFIG_RTL8169 in the board configuration
header. Now it has been fixed for a big-endian mpc8241 based
linkstation platform. This patch presents the necessary endianness
conversion fixes.
Signed-off-by: Guennadi Liakhovetski <lg@denx.de>
