Attempt to handle cases with a downstream port of a PCIe switch where
link training never completes and the link continues switching between
speeds indefinitely with the data link layer never reaching the active
state.
It has been observed with a downstream port of the ASMedia ASM2824 Gen 3
switch wired to the upstream port of the Pericom PI7C9X2G304 Gen 2
switch, using a Delock Riser Card PCI Express x1 > 2 x PCIe x1 device,
P/N 41433, wired to a SiFive HiFive Unmatched board. In this setup the
switches are supposed to negotiate the link speed of preferably 5.0GT/s,
falling back to 2.5GT/s.
However the link continues oscillating between the two speeds, at the
rate of 34-35 times per second, with link training reported repeatedly
active ~84% of the time, e.g.:
02:03.0 PCI bridge [0604]: ASMedia Technology Inc. ASM2824 PCIe Gen3 Packet Switch [1b21:2824] (rev 01) (prog-if 00 [Normal decode])
[...]
Bus: primary=02, secondary=05, subordinate=05, sec-latency=0
[...]
Capabilities: [80] Express (v2) Downstream Port (Slot+), MSI 00
[...]
LnkSta: Speed 5GT/s (downgraded), Width x1 (ok)
TrErr- Train+ SlotClk+ DLActive- BWMgmt+ ABWMgmt-
[...]
LnkCtl2: Target Link Speed: 8GT/s, EnterCompliance- SpeedDis+, Selectable De-emphasis: -3.5dB
Transmit Margin: Normal Operating Range, EnterModifiedCompliance- ComplianceSOS-
Compliance De-emphasis: -6dB
[...]
Forcibly limiting the target link speed to 2.5GT/s with the upstream
ASM2824 device makes the two switches communicate correctly however:
02:03.0 PCI bridge [0604]: ASMedia Technology Inc. ASM2824 PCIe Gen3 Packet Switch [1b21:2824] (rev 01) (prog-if 00 [Normal decode])
[...]
Bus: primary=02, secondary=05, subordinate=09, sec-latency=0
[...]
Capabilities: [80] Express (v2) Downstream Port (Slot+), MSI 00
[...]
LnkSta: Speed 2.5GT/s (downgraded), Width x1 (ok)
TrErr- Train- SlotClk+ DLActive+ BWMgmt- ABWMgmt-
[...]
LnkCtl2: Target Link Speed: 2.5GT/s, EnterCompliance- SpeedDis+, Selectable De-emphasis: -3.5dB
Transmit Margin: Normal Operating Range, EnterModifiedCompliance- ComplianceSOS-
Compliance De-emphasis: -6dB
[...]
and then:
05:00.0 PCI bridge [0604]: Pericom Semiconductor PI7C9X2G304 EL/SL PCIe2 3-Port/4-Lane Packet Switch [12d8:2304] (rev 05) (prog-if 00 [Normal decode])
[...]
Bus: primary=05, secondary=06, subordinate=09, sec-latency=0
[...]
Capabilities: [c0] Express (v2) Upstream Port, MSI 00
[...]
LnkSta: Speed 2.5GT/s (downgraded), Width x1 (downgraded)
TrErr- Train- SlotClk+ DLActive- BWMgmt- ABWMgmt-
[...]
LnkCtl2: Target Link Speed: 5GT/s, EnterCompliance- SpeedDis-
Transmit Margin: Normal Operating Range, EnterModifiedCompliance- ComplianceSOS-
Compliance De-emphasis: -6dB
[...]
Make use of this observation then and attempt to detect the inability to
negotiate the link speed automatically, and then handle it by hand. Use
the Data Link Layer Link Active status flag as the primary indicator of
successful link speed negotiation, but given that the flag is optional
by hardware to implement (the ASM2824 does have it though), resort to
checking for the mandatory Link Bandwidth Management Status flag showing
that the link speed or width has been changed in an attempt to correct
unreliable link operation (the ASM2824 does set it too).
If these checks indicate that link may not operate correctly, then poll
the Data Link Layer Link Active status flag along with the Link Training
flag for the duration of 200ms to see if the link has stabilised, that
is either that the Data Link Layer Link Active status flag has been set
or that Link Training has been inactive during at least the second half
of the interval.
If that has indicated failure, restrict the target speed to 2.5GT/s,
request a link retrain and check again if the link has stabilised. If
that does not work either, then restore the original speed setting and
claim defeat, otherwise we are done.
NB interestingly enough with the ASM2824 vs PI7C9X2G304 configuration
referred above asking the ASM2824 to retrain with a higher target link
speed once the 2.5GT/s speed has been negotiated makes the two devices
successfully negotiate 5.0GT/s. Lifting the 2.5GT/s speed restriction
would however prevent our workaround from working with an OS that issues
a reset and that is unaware of the problem. This is because the devices
would then try to negotiate a higher link speed from scratch and fail,
while the sticky property of the Target Link Speed setting will keep the
2.5GT/s speed restriction across a reset.
Keep the 2.5GT/s speed restriction then, conservatively, if functional
once applied.
Signed-off-by: Maciej W. Rozycki <macro@orcam.me.uk>
Reviewed-by: Stefan Roese <sr@denx.de>
It is common to set all base address bits to one and all limit address bits
to zero for disabling address forwarding. Forwarding is disabled when base
address is higher than limit address, so this change should not have any
effect.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
If U-Boot does not have any I/O resource for assignment then disable I/O
forwarding in PCI bridge autoconfiguration code. Default initial state of
PCI bridge IO registers is unspecified, therefore they can be in enabled if
U-Boot does not touch them.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
Function dm_pciauto_prescan_setup_bridge() configures base address
registers, therefore it should read type of IO from base address registers
(and not from limit address registers).
Note that base and limit address registers should have same type, so this
change is just usage correction and has no functional change on correctly
working hardware.
Fixes: 8e85f36a8f ("pci: Fix configuring io/memory base and limit registers of PCI bridges")
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
PCI Rom Address is currently supported only for Normal (0x00) and
Bridge (0x01) header types. Fix code accordingly.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
To avoid a build warning with W=1, provide a function prototype for
dm_pciauto_prescan_setup_bridge, which is a non-static function whose
definition is inside pci_auto.c.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Lower 4 bits of PCI_MEMORY_BASE and PCI_MEMORY_LIMIT registers are reserved
and should be zero. So do not set them to non-zero value.
Lower 4 bits of PCI_PREF_MEMORY_BASE and PCI_PREF_MEMORY_LIMIT registers
contain information if 64-bit memory addressing is supported. So preserve
this information when overwriting these registers.
Lower 4 bits of PCI_IO_BASE and PCI_IO_LIMIT register contain information
if 32-bit io addressing is supported. So preserve this information and do
not try to configure 32-bit io addressing (via PCI_IO_BASE_UPPER16 and
PCI_IO_LIMIT_UPPER16 registers) when it is unsupported.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
These boards have not been converted to CONFIG_DM_PCI by the deadline
and is also missing conversion to CONFIG_DM. Remove them.
Cc: Reinhard Arlt <reinhard.arlt@esd-electronics.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
This patch completely removes CONFIG_PCI_ENUM_ONLY from the PCI code as
it is not configured for any board (any more). With this removal, some
PCI related files get cleaned up a bit.
Additional, dm_pciauto_setup_device() is now static, as it's not
referenced from any code outside of this C file.
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Simon Glass <sjg@chromium.org>
Cc: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
On my DS414, some PCI devices return odd values when probing BAR sizes.
An obvious case is all-ones response, the Linux driver
(drivers/pci/probe.c) catches those explicitly and a comment explains
that either bit 0 or bit 1 must be clear (depending on MEM or IO type).
Other BARs return e.g. 0xfff0000f or 0xfff00004 and thus manage to break
size calculation due to the "middle" zeroes. Mitigate that copying more
or less what Linux does and do a "find least bit set".
Signed-off-by: Phil Sutter <phil@nwl.cc>
Reviewed-by: Stefan Roese <sr@denx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Now that we know the sequence number at bind time, there is no need for
special-case code in dm_pci_hose_probe_bus().
Note: the PCI_CAP_ID_EA code may need a look, but there are no test
failures so I have left it as is.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present various drivers etc. access the device's 'seq' member directly.
This makes it harder to change the meaning of that member. Change access
to go through a function instead.
The drivers/i2c/lpc32xx_i2c.c file is left unchanged for now.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present this fails silently which can be confusing since some devices
on the PCI bus may not work correctly. Show a message in this case.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
[bmeng: add a '\n' in the PCI autoconfig fail message]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
vme8349.h contains two separate boards: The vme8349 itself, and the
caddy2 board. The caddy2 board is chosen by setting certain config
variables. Create a proper config file for the caddy2 board to make
Kconfig migration easier.
Furthermore, simplify the vme8349 and caddy2 configs by keeping only the
options necessary for each board.
Signed-off-by: Mario Six <mario.six@gdsys.cc>
CONFIG_TARGET_VME8349 can replace CONFIG_VME8349. Hence, replace
CONFIG_VME8349 with CONFIG_TARGET_VME8349, and remove CONFIG_VME8349.
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Currently, if we happen to allocate an address requiring 64 bits to a
device only supporting 32-bit BARs, the address eventually gets silently
truncated to 32 bits. Avoid this by adding a new flag to
pciauto_region_allocate() to bail out in such situations.
Signed-off-by: Tuomas Tynkkynen <tuomas.tynkkynen@iki.fi>
Reviewed-by: Simon Glass <sjg@chromium.org>
When U-Boot started using SPDX tags we were among the early adopters and
there weren't a lot of other examples to borrow from. So we picked the
area of the file that usually had a full license text and replaced it
with an appropriate SPDX-License-Identifier: entry. Since then, the
Linux Kernel has adopted SPDX tags and they place it as the very first
line in a file (except where shebangs are used, then it's second line)
and with slightly different comment styles than us.
In part due to community overlap, in part due to better tag visibility
and in part for other minor reasons, switch over to that style.
This commit changes all instances where we have a single declared
license in the tag as both the before and after are identical in tag
contents. There's also a few places where I found we did not have a tag
and have introduced one.
Signed-off-by: Tom Rini <trini@konsulko.com>
This patch is to change U-Boot PCI bus assignement compliant with Linux.
It means each PCIe controller's bus number is 0, not the current maximum
PCI bus number, when start to scan this controller.
Signed-off-by: Minghuan Lian <Minghuan.Lian@nxp.com>
Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Fix the following compiler warnings when DEBUG is on.
warning: 'bar_res' may be used uninitialized in this function.
drivers/pci/pci_auto.c:101:21:
if (!enum_only && pciauto_region_allocate(bar_res, bar_size,
^
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
This function is only available for compatibility with old code. Avoid
using it in the uclass.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
For this class it is intended to set up the PCI device, so add a comment to
indicate this. This avoids a coverity warning.
Reported-by: Coverity (CID: 134194)
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
At present we are using legacy functions even in the auto-configuration code
used by driver model. Add a new pci_auto.c version which uses the correct
API.
Create a new pci_internal.h header to hold functions that are used within
the PCI subsystem, but are not exported to other drivers.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
This file should not be used with driver model as it has lots of legacy/
compatibility functions. Rename it to make this clear.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
PCI_COMMAND_IO bit must be set for VGA device as it needs to respond
to legacy VGA IO address.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
PCI_HEADER_TYPE register (offset 0x0e) bit 7 is an indicator
for multi-function devices. We should mask it off before using
it as the header type.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
At present the PCI output displays 'Mem' when it allocates memory for a PCI
device, whether it is prefetchable or not. There is a distinction since the
memory comes from separate pools. Use 'Prf' instead of 'Mem' when allocating
prefetchable memory.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This option is not used by any board but appears to still be useful, at least
for testing. With recent commits it does not build, so fix it.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This file does not need its own way of doing debug(). Clean it up to use the
new way.
Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Pavel Machek <pavel@denx.de>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Commit aec241d "dm: pci: Use the correct hose when configuring devices"
was an attempt to fix pci bridge device configuration, but unfortunately
that does not work 100%. In pciauto_config_devices(), the fix tried to
call pciauto_config_device() with a ctlr_hose which is supposed to be
the root controller hose, however when walking through a pci topology
with 2 or more pci bridges this logic simply fails.
The call chain is: pciauto_config_devices()->pciauto_config_device()
->dm_pci_hose_probe_bus(). Here the call to dm_pci_hose_probe_bus()
does not make any sense as the given hose is not the bridge device's
hose, instead it is either the root controller's hose (case#1: if it
is the 2nd pci bridge), or the bridge's parent bridge's hose (case#2:
if it is the 3rd pci bridge). In both cases the logic is wrong.
For example, for failing case#1 if the bridge device to config has the
same devfn as one of the devices under the root controller, the call
to pci_bus_find_devfn() will return the udevice of that pci device
under the root controller as the bus, but this is wrong as the udevice
is not a bus which does not contain all the necessary bits associated
with the udevice which causes further failures.
To correctly support pci bridge device configuration, we should still
call pciauto_config_device() with the pci bridge's hose directly.
In order to access valid pci region information, we need to refer to
the root controller simply by a call to pci_bus_to_hose(0) and get the
region information there in the pciauto_prescan_setup_bridge(),
pciauto_postscan_setup_bridge() and pciauto_config_device().
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
In driver model, each pci bridge device has its own hose structure.
hose->first_busno points to the bridge device's device number, so
we should not substract hose->first_busno before programming the
bridge device's primary/secondary/subordinate bus number registers.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Currently PCI expansion ROM address is assigned by a call to
pciauto_setup_rom() outside of the pci auto config process.
This does not work when expansion ROM is on a device behind
PCI bridge where bridge's memory limit register was already
programmed to a value that does not cover the newly assigned
expansion ROM address. To fix this, we should configure the
ROM address during the auto config process.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Simon Glass <sjg@chromium.org>
Acked-by: Simon Glass <sjg@chromium.org>
The legacy IDE device has a BAR4 (Bus Master Interface BAR) which
needs to be configured.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Acked-by: Simon Glass <sjg@chromium.org>
Remove the '#undef DEBUG' in pci_auto.c so that we can enable debug
message output via '-DDEBUG'.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
Add a uclass for PCI controllers and a generic one for PCI devices. Adjust
the 'pci' command and the existing PCI support to work with this new uclass.
Keep most of the compatibility code in a separate file so that it can be
removed one day.
TODO: Add more header file comments to the new parts of pci.h
Signed-off-by: Simon Glass <sjg@chromium.org>
PCI specification allow prefetchable memory to be 32-bit or 64-bit.
PCI express specification states that all memmory bars for prefetchable
memory must be implemented as 64-bit. They all require that 64 bit
prefetchble memory are suported especially when u-boot is ported to
more and more 64bit processors.
Signed-off-by: David Feng <fenghua@phytium.com.cn>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
U-Boot has never cared about the type when we get max/min of two
values, but Linux Kernel does. This commit gets min, max, min3, max3
macros synced with the kernel introducing type checks.
Many of references of those macros must be fixed to suppress warnings.
We have two options:
- Use min, max, min3, max3 only when the arguments have the same type
(or add casts to the arguments)
- Use min_t/max_t instead with the appropriate type for the first
argument
Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Pavel Machek <pavel@denx.de>
Acked-by: Lukasz Majewski <l.majewski@samsung.com>
Tested-by: Lukasz Majewski <l.majewski@samsung.com>
[trini: Fixup arch/blackfin/lib/string.c]
Signed-off-by: Tom Rini <trini@ti.com>
Forcibly set hose->pci_prefetch to NULL to make sure it will be setup.
This will help if for any reason callers didn't make sure themselves to
NULL the field.
Signed-off-by: Thierry Reding <treding@nvidia.com>
The following commit introduced some warnings associated with using
pci_addr_t instead of a proper 32-bit data type.
commit af778c6d9e
Author: Andrew Sharp <andywyse6@gmail.com>
Date: Wed Aug 1 12:27:16 2012 +0000
pci: fix errant data types and corresponding access functions
On some platforms pci_addr_t is defined as a 64-bit data type so its not
proper to use with pci_{read,write}_config_dword.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Introduce CONFIG_PCI_ENUM_ONLY variable for platforms that just want a
quick enumberation of the PCI devices, but don't need any setup work done.
This is very beneficial on platforms that have u-boot loaded by another
boot loader which does a more sophisticated job of setup of PCI devices
than u-boot. That way, u-boot can just read what's there and get on
with life. This is what SeaBIOS does.
Signed-off-by: Andrew Sharp <andywyse6@gmail.com>
I tried to clean up the white space and formatting offenses and
inconsistencies in the generic PCI code that obviously has been around for
some time. Emphasis on large increases in readability and maintainability
and consistency. I omitted the platform/processor specific files in
the drivers/pci directory because I wanted to leave those file to those
that care more about them.
Signed-off-by: Andrew Sharp <andywyse6@gmail.com>
In a couple of places, unsigned int and pci_config_*_dword were being
used when u16 and _word should be used. Unsigned int was also being
used in a couple of places that should be pci_addr_t.
Signed-off-by: Andrew Sharp <andywyse6@gmail.com>
As discussed on the list, move "arch/ppc" to "arch/powerpc" to
better match the Linux directory structure.
Please note that this patch also changes the "ppc" target in
MAKEALL to "powerpc" to match this new infrastructure. But "ppc"
is kept as an alias for now, to not break compatibility with
scripts using this name.
Signed-off-by: Stefan Roese <sr@denx.de>
Acked-by: Wolfgang Denk <wd@denx.de>
Acked-by: Detlev Zundel <dzu@denx.de>
Acked-by: Kim Phillips <kim.phillips@freescale.com>
Cc: Peter Tyser <ptyser@xes-inc.com>
Cc: Anatolij Gustschin <agust@denx.de>
This patch adds support for the esd VME8349 board equipped with the
MPC8349. It's a VME PMC carrier board equipped with the Tundra
TSI148 VME-bridge.
Signed-off-by: Reinhard Arlt <reinhard.arlt@esd-electronics.com>
Signed-off-by: Stefan Roese <sr@denx.de>
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Use the standard lowercase "x" capitalization that other Freescale
architectures use for CPU defines to prevent confusion and errors
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
