The link_startup_again logic was added in Linux to handle device
that were set in LinkDown state, which should not be the case since U-boot
doesn't set LinkDown state are init, and Linux sets the device active
in ufshcd_init() for the first link startup.
ufshcd_set_ufs_dev_active(hba) is called at ufshcd_init() right before
scheduling an ufshcd_async_scan that will call ufshcd_device_init() then ufshcd_link_startup().
The comment in probe says:
/*
* We are assuming that device wasn't put in sleep/power-down
* state exclusively during the boot stage before kernel.
* This assumption helps avoid doing link startup twice during
* ufshcd_probe_hba().
*/
we can assume the same from U-Boot.
While it worked to far, it breaks link startup for Qualcomm Controllers v5,
let's just remove the logic.
Tested-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@amd.com>
Tested-by: Julius Lehmann <lehmanju@devpi.de>
Link: https://lore.kernel.org/r/20240930-topic-ufs-enhancements-v3-12-58234f84ab89@linaro.org
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
There is no need to flush and invalidate all data updated by the
driver, mainly because on ARM platforms flush also invalidates
the cachelines.
Split the function in two and add the appropriate cacheline
invalidates after the UFS DMA operation finishes to make sure
we read from memory.
Flushing then invalidating cacheline unaligned data causes data
corruption issues on Qualcomm platforms, and is largely unnecessary
anyway, so let's cleanup the cache operations.
Reviewed-by: Neha Malcom Francis <n-francis@ti.com>
Tested-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@amd.com>
Tested-by: Julius Lehmann <lehmanju@devpi.de>
Link: https://lore.kernel.org/r/20240930-topic-ufs-enhancements-v3-3-58234f84ab89@linaro.org
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
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>
Extend the version check to handle UFS 3.1 controllers as well.
Tested on QEMU emulated UFS 3.1 controller.
Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Neha Malcom Francis <n-francis@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
This adds a simple PCI based UFS controller driver with a QEMU
emulated UFS controller on the PCI bus.
Requiring QEMU v8.2+.
Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Neha Malcom Francis <n-francis@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Check if the UFS controller is on the PCI bus, and get its register
base address accordingly.
Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Neha Malcom Francis <n-francis@ti.com>
Add a line feed to improve readability of some dev_xxx() messages.
Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Neha Malcom Francis <n-francis@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
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 "#include <common.h>" is being phased out in favor of more fine
grained header management, i.e. ideally include a subset of headers
that are really needed. Remove it from this driver.
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
Add support for Renesas R-Car UFS controller which needs vendor-specific
initialization.
Ported from Linux kernel as of commit
c2ab666072bc ("scsi: ufs: Explicitly include correct DT includes")
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
Add function to flush and invalidate cache over request and response
queue entries, and perform flush and optional invalidate over block
layer data that are passed into the UFS layer. This makes it possible
to use UFS with caches enabled.
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Reviewed-by: Bhupesh Sharma <bhupesh.sharma@linaro.org>
Tested-by: Bhupesh Sharma <bhupesh.sharma@linaro.org>
Use utp_transfer_req_desc pointer to reference to utrdl queue
instead of referencing the queue directly. This makes the code
more consistent. No functional change.
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Reviewed-by: Bhupesh Sharma <bhupesh.sharma@linaro.org>
Tested-by: Bhupesh Sharma <bhupesh.sharma@linaro.org>
Pass the hba pointer itself to ufshcd_prepare_req_desc_hdr()
instead of duplicating utp_transfer_req_desc access at each
call site. No functional change.
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Reviewed-by: Bhupesh Sharma <bhupesh.sharma@linaro.org>
Tested-by: Bhupesh Sharma <bhupesh.sharma@linaro.org>
Add UFSHCD_QUIRK_HIBERN_FASTAUTO quirk for host controllers which supports
auto-hibernate the capability but only FASTAUTO mode.
Ported from Linux kernel commit
2f11bbc2c7f3 ("scsi: ufs: core: Add UFSHCD_QUIRK_HIBERN_FASTAUTO")
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Reviewed-by: Bhupesh Sharma <bhupesh.sharma@linaro.org>
Tested-by: Bhupesh Sharma <bhupesh.sharma@linaro.org>
Add UFSHCD_QUIRK_BROKEN_64BIT_ADDRESS for host controllers which do not
support 64-bit addressing.
Ported from Linux kernel commit
6554400d6f66 ("scsi: ufs: core: Add UFSHCD_QUIRK_BROKEN_64BIT_ADDRESS")
with ufs_scsi_buffer_aligned() based on U-Boot generic bounce buffer.
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Reviewed-by: Bhupesh Sharma <bhupesh.sharma@linaro.org>
Tested-by: Bhupesh Sharma <bhupesh.sharma@linaro.org>
Instead of using the hard-coded value of 0x1f, use 'TASK_TAG'
macro instead to construct the ucd_req_ptr->header.dword_0
This is in sync with what the Linux UFS driver does, i.e.
set the byte0 equal to TASK_TAG (see [1]).
Setting it to a fixed value of 0x1f is wrong as we define
TASK_TAG as 0 inside u-boot ufs framework. So, instead we
should use the macro value directly.
[1]. https://github.com/torvalds/linux/blob/master/drivers/ufs/core/ufshcd.c#L2705
Signed-off-by: Bhupesh Sharma <bhupesh.sharma@linaro.org>
The fdt_addr_t and phys_addr_t size have been decoupled. A 32bit CPU
can expect 64-bit data from the device tree parser, so use
dev_read_addr_ptr instead of the dev_read_addr function in the
various files in the drivers directory that cast to a pointer.
As we are there also streamline the error response to -EINVAL on return.
Signed-off-by: Johan Jonker <jbx6244@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
