Jiaxun Yang <jiaxun.yang@flygoat.com> says:
This is a huge series which promoted MIPS/Boston target into a
usable state, with fixes to drivers and general framework issues
I found in this process.
I also converted the target to OF_UPSTREAM.
This target is covered by QEMU, to test on QEMU:
```
make boston64r6el_defconfig
make
qemu-system-mips64el -M boston -cpu I6500 -bios ./u-boot.bin -nographic
```
Link: https://lore.kernel.org/r/20240517-boston-v3-0-1ea7d23f4a1d@flygoat.com
Ensure that we are using correct physical/virtual address for
DMA buffer write and hardware register settings.
The convention is: in ahci_ioports all pointers are virtual,
that will be converted to physical address when writing to
hardware registers or into sg/cmd_tbl.
Also fixed 64bit physical address support for dwc_ahsata, ensure
higher bits are written into registers/sg properly.
Use memalign for allocating aligned buffer in dwc_ahsata so we
don't have to do our own alignment in driver.
Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Caleb Connolly <caleb.connolly@linaro.org> says:
SCSI devices like UFS may maintain their own cache to speed up writes,
however this is lost on board reset (and may be lost on device removal
or reset by OS drivers).
Currently this can be worked around by "waiting for a while" after
writing data to disk, but of course this is not an acceptable solution.
Ideally U-Boot would have a mechanism to flush caches during board
reset, but until that logic is hooked up let's be sure that all writes
are actually propagated to the storage device so that we don't lose data
on board reset.
The same logic was already implemented just for the AHCI backend, this
duplicated logic has been removed and support for the SYNC_CACHE command
is added to AHCI.
This is particularly noticeable during capsule updates, since the update
file is deleted and the board is reset immediately afterwards which
resulted in the same capsule update being applied over and over again.
This specifically fixes Qualcomm SDM845 devices with UFS 2.1, but likely
all UFS devices that use a cache.
Link: https://lore.kernel.org/r/20250326-scsi-sync-on-write-v2-0-12ab05bd464b@linaro.org
The SCSI layer now issues a SYNC_CACHE command after every write to
ensure there is no data loss due to a board reset after write.
Implement support for this command and remove the same logic from the
ATA write path to be consistent with other SCSI backends.
Ranges are not supported and the whole cache will be flushed in all
cases.
This was done per iteration in ata_scsiop_read_write(), but it's not
clear why this was the case, calling it once for the entire write ought
to achieve the same result.
Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
In the case of a memory allocation error, the ahci_port_start() function
tries to free the `pp' pointer.
This pointer was not dynamically allocated but does in fact point to an
element of the port[] array member of the struct ahci_uc_priv.
Remove the erroneous call to free() to fix this.
Fixes: 4782ac80b0 ("Add AHCI support to u-boot")
Signed-off-by: Vincent Stehlé <vincent.stehle@arm.com>
Cc: Tom Rini <trini@konsulko.com>
Cc: Jason Jin <jason.jin@freescale.com>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
As part of bringing the master branch back in to next, we need to allow
for all of these changes to exist here.
Reported-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Tom Rini <trini@konsulko.com>
When bringing in the series 'arm: dts: am62-beagleplay: Fix Beagleplay
Ethernet"' I failed to notice that b4 noticed it was based on next and
so took that as the base commit and merged that part of next to master.
This reverts commit c8ffd1356d, reversing
changes made to 2ee6f3a5f7.
Reported-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Tom Rini <trini@konsulko.com>
Remove <common.h> from this driver directory and when needed
add missing include files directly.
Reviewed-by: Tony Dinh <mibodhi@gmail.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
The migration deadline for moving to DM_SCSI was v2023.04. A further
reminder was sent out in August 2023 to the remaining platforms that had
not migrated already, and that a few more over the line (or configs
deleted).
With this commit we:
- Rename CONFIG_DM_SCSI to CONFIG_SCSI.
- Remove all of the non-DM SCSI code. This includes removing other
legacy symbols and code and removes some legacy non-DM AHCI code.
- Some platforms that had previously been DM_SCSI=y && SCSI=n are now
fully migrated to DM_SCSI as a few corner cases in the code assumed
DM_SCSI=y meant SCSI=y.
Signed-off-by: Tom Rini <trini@konsulko.com>
The non DM code path already would enable pci bus mastering. Do the
same for the DM code path.
Fixes AHCI problems I am seeing on an Intel Apollolake device.
Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
[trini: Use ahci_dev not dev in the calls]
Signed-off-by: Tom Rini <trini@konsulko.com>
CONFIG_SYS_SCSI_MAX_SCSI_ID is not defined if CONFIG_SCSI=n.
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The controller indicates the number of ports but also has a port map
which specifies which ports are actually valid. Make use of this to
avoid trying to send commands to an invalid port.
This avoids a crash on some controllers.
Signed-off-by: Simon Glass <sjg@chromium.org>
Add a mask parameter to control the lookup of the PCI region from which
the mapping can be made.
Signed-off-by: Andrew Scull <ascull@google.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Evolve dm_pci_map_bar() to include an offset and length parameter. These
allow a portion of the memory to be mapped and range checks to be
applied.
Passing both the offset and length as zero results in the previous
behaviour and this is used to migrate the previous callers.
Signed-off-by: Andrew Scull <ascull@google.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
When building highbank_defconfig with LTO, the compiler complains about
type mismatch of function ahci_link_up().
The third parameter of this function is of type u8 in
drivers/ata/ahci.c, but of type int in board/highbank/ahci.c.
There is no reason in using u8, and the code using this function
actually passes an int variable into the function (so it is implicitly
converted to u8).
Change the type of this parameter to int in drivers/ata/ahci.c.
Signed-off-by: Marek Behún <marek.behun@nic.cz>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This patch adds a few missing virt_to_phys() to use the correct physical
address for DMA operations in the common AHCI code. This is done to
support the big-endian MIPS Octeon platform.
Additionally the code a cleaned up a bit (remove some empty lines) and
made a bit better readable.
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Simon Glass <sjg@chromium.org>
Cc: Aaron Williams <awilliams@marvell.com>
Cc: Chandrakala Chavva <cchavva@marvell.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
After check for maximum between max id and available ports, also check
if available port count is less than max id and update.
In the case of the CN8030 OcteonTX SoC max_id needs to be reduced to
the number of ports found otherwise the following occurs on a scan:
GW6404-B> scsi scan
scanning bus for devices...
Target spinup took 0 ms.
AHCI 0001.0300 32 slots 1 ports 6 Gbps 0x1 impl SATA mode
flags: 64bit ncq ilck stag pm led clo only pmp fbss pio slum part ccc
apst
Device 0: (0:0) Vendor: ATA Prod.: SanDisk SD8SFAT0 Rev: Z233
Type: Hard Disk
Capacity: 61057.3 MB = 59.6 GB (125045424 x 512)
"Synchronous Abort" handler, esr 0x96000006
elr: 000000000052f824 lr : 000000000052fa10 (reloc)
elr: 000000007fee9824 lr : 000000007fee9a10
x0 : 0000000000000001 x1 : 0000000000000001
x2 : 000000007bea3528 x3 : 000000007bea3580
x4 : 0000000000000200 x5 : 0000000000000000
x6 : 0000000000000002 x7 : 000000007bea3540
x8 : 00000000fffffff8 x9 : 0000000000000008
x10: 00000000000186a0 x11: 000000000000000d
x12: 0000000000000006 x13: 000000000001869f
x14: 0000000000000007 x15: 00000000ffffffff
x16: 000000007ff439a5 x17: 000000007ff5730c
x18: 000000007bea9de0 x19: 000000007ff7a580
x20: 000000007bec79f8 x21: 0000000000000000
x22: 000000007bea3580 x23: 0000000000000000
x24: 0000000000000000 x25: 000000007bec7a00
x26: 00000000ffffffc0 x27: 000000007bec79d0
x28: 000000007beb51c0 x29: 000000007bea3480
Code: 91246800 940130c2 12800000 1400004f (b9402ae0)
Resetting CPU ...
Signed-off-by: Suneel Garapati <sgarapati@marvell.com>
Reviewed-by: Stefan Roese <sr@denx.de>
We use 'priv' for private data but often use 'platdata' for platform data.
We can't really use 'pdata' since that is ambiguous (it could mean private
or platform data).
Rename some of the latter variables to end with 'plat' for consistency.
Signed-off-by: Simon Glass <sjg@chromium.org>
For SATA controller found on OcteonTX SoC's, use non-standard PCI BAR0
instead of BAR5.
Signed-off-by: Suneel Garapati <sgarapati@marvell.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
These functions are CPU-related and do not use driver model. Move them to
cpu_func.h
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
In 32bits mode upper bits need to be set to 0, otherwise controller will
try to DMA into not existing memory and stops with error.
Tested-by: Frank Wunderlich <frank-w@public-files.de>
Signed-off-by: Frank Wunderlich <frank-w@public-files.de>
Signed-off-by: Oleksandr Rybalko <ray@ddteam.net>
The scsi_scan_dev() is looping over the number of uc_plat->max_id.
The number of actual ports a AHCI controller has can be greater than
max_id. Update uc_plat->max_id to make SCSI scan all detected ports.
Signed-off-by: Aiden Park <aiden.park@intel.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
malloc(..) and memalign(..) are both allocating memory and as a result
we leak the memory allocated with malloc(..).
Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
When using device model this sort of hardcoded limits aren't used or
necessary.
Signed-off-by: Tuomas Tynkkynen <tuomas.tynkkynen@iki.fi>
Reviewed-by: Simon Glass <sjg@chromium.org>
The loop in ahci_start_ports() is looping over the maximum number of
SCSI devices in the system, which can be larger than the amount of ports
a particular AHCI controller has. The extra looping isn't directly
harmful because the link_port_map bitmap won't have the bit set for a
nonexistent port, but it is wasteful. Replace the loop limit with the
port count of the AHCI controller instead.
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 is expected to be attached to the uclass and the code operates that
way, but the uclass has not been updated. Fix it to avoid using memory at
address 0.
Signed-off-by: Simon Glass <sjg@chromium.org>
Fixes: 47fc61a (dm: ahci: Drop use of probe_ent)
At present the AHCI SCSI driver only supports PCI with driver model.
Rename the existing function to indicate this and add support for adding
a non-PCI controller .
Signed-off-by: Simon Glass <sjg@chromium.org>
Running 'scsi scan' command causes scsi_scan_dev() to be called,
from which device_probe() is called and consequently AHCI driver
probe routine will be called as SCSI driver's parent, and finally
ahci_probe_scsi() calls scsi_scan_dev() again.
Remove the call to scsi_scan_dev() in ahci_probe_scsi().
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Some AHCI drivers use SCSI under the hood. Rather than making the AHCI
driver be in the SCSI uclass it makes sense to have the AHCI device create
a SCSI device as a child. That way we can handle any AHCI-specific
operations rather than trying to pretend tha the device is just SCSI.
To handle this we need to provide a way for AHCI drivers to bind a SCSI
device as its child, and probe it. Add functions for this.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Change this function to return an error number instead of true/false.
This allows us to return a proper error number.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
The SCSI uclass currently has no operations. It just uses the global SCSI
functions. Fix this by adding operations to the only two drivers that use
the uclass, and replacing the global functions with those defined locally
in the SCSI code.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
With driver model we need to define implementations of exec() and
bus_reset() separately for each SCSI driver. As a first step, create a
local version of each function in the AHCI driver and call each from its
global version.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
With driver model these functions need a device pointer. Add one even
when CONFIG_DM_SCSI is not defined. This avoids having ugly conditional
function prototypes, When CONFIG_DM_SCSI is not defined we can just ignore
the pointer.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
With driver model we cannot have static data or assume that there is only
one device of each time. Adjust the code so that 'probe_ent' is not needed
with driver model. Add a new ahci_init_dm() function which can init AHCI
for driver model without re-allocating the uclass data. Move over the only
existing driver to use this new function.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Two AHCI drivers use SCSI with CONFIG_DM_SCSI. The SCSI uclass calls
scsi_low_level_init() which is implemented by ahci.c. If
CONFIG_SCSI_AHCI_PLAT is defined it does one thing and if it is not
it does something else.
We don't need to call through scsi_low_level_init() to get the init
completed. Instead, adjust the two drivers to call into AHCI directly.
Drop the post-probe init in the SCSI uclass. This means that driver model
doesn't need to use scsi_low_level_init(). It is a legacy function and
driver model should use a driver's probe() method instead.
While we are here, add a comment to the top of the file explaining what
ahci.c does.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This code is duplicated. Create a ahci_start_ports() function to handle
this and call it from both places.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
With driver model we need each device to have its own state. As a step
towards this, restrict use of the global 'probe_ent' to just a few places
in the file. This will allow us to add driver-model functions which can
pass the correct data around.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This array relates to the AHCI controller so should be exist out on its
own in the file. Move it into the structure. Adjust functions that need
access to this to take the structure as a parameter.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This is not a very useful name since once it is probed it still hangs
around. With driver model we will use uclass data for this, so rename the
struct.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
