add block storage emulation for NVM XIP flash devices
Some paltforms such as Corstone-1000 need to see NVM XIP raw flash
as a block storage device with read only capability.
Here NVM flash devices are devices with addressable
memory (e.g: QSPI NOR flash).
The implementation is generic and can be used by different platforms.
Two drivers are provided as follows.
nvmxip-blk :
a generic block driver allowing to read from the XIP flash
nvmxip Uclass driver :
When a device is described in the DT and associated with
UCLASS_NVMXIP, the Uclass creates a block device and binds it with
the nvmxip-blk.
Platforms can use multiple NVM XIP devices at the same time by defining a
DT node for each one of them.
Signed-off-by: Abdellatif El Khlifi <abdellatif.elkhlifi@arm.com>
To quote the author:
This code was converted to driver model a long time again but it was a
pretty rough conversion. It introduced a few minor bugs, e.g. the device
capacity is incorrect and some flags are lost (such as lba48).
This series tidies up the code and fixes these bugs. This involves quite
a bit of refactoring, so it is done one patch at a time for easier
review.
The code has quite a few unnecessary brackets and comparisons to zero,
etc. Fix these up as well as some upper-case hex values and use of 0x in
printf() strings.
Signed-off-by: Simon Glass <sjg@chromium.org>
Avoid the use of the function name in a few of the debug() calls, since
this causes a checkpatch warning. Convert all other calls too.
Use lower-case hex consistently.
Signed-off-by: Simon Glass <sjg@chromium.org>
Fix a longstanding bug where the LBA is calculated as the size of the
media instead of the number of blocks. This was perhaps not noticed
earlier since it prints the correct value first, before setting the wrong
value.
Drop the unnecessary blksz variable while we are here.
Signed-off-by: Simon Glass <sjg@chromium.org>
Fixes: 68e6f221ed ("block: ide: Fix block read/write with driver model")
We only use one member of the ide_dev_desc[] array at a time and it does
not stick around outside ide_probe(). Use a single element instead.
Copy over the missing members of blk_desc at the same, since this was
missing from the previous code.
Signed-off-by: Simon Glass <sjg@chromium.org>
Fixes: 68e6f221ed ("block: ide: Fix block read/write with driver model")
Rather than having the caller fill some of this in, do it all in the
ide_ident() function, since it knows all the values.
Signed-off-by: Simon Glass <sjg@chromium.org>
Update ide_ident() to indicate whether it finds a device or not. Use
that to decide whether to create a block device for it, rather than
looking DEV_TYPE_UNKNOWN.
Signed-off-by: Simon Glass <sjg@chromium.org>
Most of the code uses 'desc' as the variable name for a blk descriptor.
Change ide to do the same.
Tidy up some extra brackets and types while we are here.
Leave the code in ide_probe() alone since it is about to be refactored.
Signed-off-by: Simon Glass <sjg@chromium.org>
The two loops in this function operate on the same ide_dev_desc[] array.
Combine them to reduce duplication.
Signed-off-by: Simon Glass <sjg@chromium.org>
Now that ide_probe() is the only caller of ide_init(), move all the code
into the probe function, so it is easier to refactor it.
Move ide_dev_desc[] into ide_probe() to, since it is the only user.
Signed-off-by: Simon Glass <sjg@chromium.org>
The current implementation adds this information in the block device's
probe() function, which is called in the blk_probe_or_unbind() in
ide_probe().
It is simpler to do this in ide_probe() itself, since the effect is the
same. This helps to consolidate use of ide_dev_desc[] which we would like
to remove.
Use strlcpy() to keep checkpatch happy.
Signed-off-by: Simon Glass <sjg@chromium.org>
Use IS_ENABLED() instead for all conditions. Add the 'lba48' flag into
struct blk_desc always, since it uses very little space. Use a bool so
the meaning is clearer.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Mattijs Korpershoek <mkorpershoek@baylibre.com>
Change the if() to remove extra brackets and check for the positive case
first, i.e. when a device is found. Exit the loop in that case, with the
retry logic in the 'else' part.
Signed-off-by: Simon Glass <sjg@chromium.org>
This code is hard to follow as it uses #ifdef in a strange way. Adjust
it to avoid the preprocessor. Drop the special return for the non-ATAPI
case since we can rely on tries becoming 0 and exiting the loop.
Signed-off-by: Simon Glass <sjg@chromium.org>
Use a 'tries' variable which starts at the number of tries we want to do,
rather than a 'retries' one that stops at either 1 or 2. This will make it
easier to refactor the code to avoid the horrible #ifdefs
Signed-off-by: Simon Glass <sjg@chromium.org>
The use of atapi_read() was incorrect dropped. Fix this so that it will
be used when needed. Use a udevice for the first argument of atapi_read()
so it is consistent with ide_read().
This requires much of the ATAPI code to be brought out from behind the
existing #ifdef. It will still be removed by the compiler if it is not
needed.
Add an atapi flag to struct blk_desc so the information can be retained.
Fixes: 145df842b4 ("dm: ide: Add support for driver-model block devices")
Fixes: d0075059e4 ("ide: Drop non-DM code for BLK")
Reviewed-by: Mattijs Korpershoek <mkorpershoek@baylibre.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
These are not used from outside this file anymore. Make them static and
remove them from the header file.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present the code does ide_init() as a separate operation, then calls
device_probe() to copy over the information. We can call ide_init() from
probe just as easily.
The only difference is that using 'ide init' twice will do nothing.
However it already fails to copy over the new data in that case, so the
effect is the same. For now, unbind the block devices and remove the IDE
device, which causes the bus to be probed again. Later patches will fix
this up fully, so that all blk_desc data is copied across.
Since ide_reset() is only called from ide_init(), there is no need to init
the ide_dev_desc[] array. This is already done at the end of ide_init() so
drop this code.
The call to uclass_first_device() is now within the probe() function of
the same device, so does nothing. Drop it.
Signed-off-by: Simon Glass <sjg@chromium.org>
It doesn't really matter if we mention things which are not present. For
example, if 'nvme' is included but the board does not support it, it just
continues with the next item in the list.
It is simpler to use the same target list for all boards, so drop the
different one for rk3399.
Signed-off-by: Simon Glass <sjg@chromium.org>
When loading a file from a block device, efiload_read_file
was using the seq_num of the device (e.g "35" of virtio_blk#35)
instead of the block device id (e.g what you get from running
the corresponding device scan command, like "virtio 0")
This cause EFI booting from these devices to fail as an
invalid device number is passed to blk_get_device_part_str:
Scanning bootdev 'virtio-blk#35.bootdev':
distro_efi_read_bootflow_file start (efi,fname=<NULL>)
distro_efi_read_bootflow_file start (efi,fname=<NULL>)
setting bootdev virtio, 35, efi/boot/bootaa64.efi, 00000000beef9a40, 170800
efi_dp_from_name calling blk_get_device_part_str
dev=virtio devnr=35 path=efi/boot/bootaa64.efi
blk_get_device_part_str (virtio,35)
blk_get_device_by_str (virtio, 35)
** Bad device specification virtio 35 **
Using default device tree: dtb/qemu-arm.dtb
No device tree available
0 efi ready virtio 1 virtio-blk#35.bootdev.par efi/boot/bootaa64.efi
** Booting bootflow 'virtio-blk#35.bootdev.part_1' with efi
blk_get_device_part_str (virtio,0:1)
blk_get_device_by_str (virtio, 0)
No UEFI binary known at beef9a40 (image buf=00000000beef9a40,addr=0000000000000000)
Boot failed (err=-22)
Signed-off-by: Mathew McBride <matt@traverse.com.au>
Signed-off-by: Simon Glass <sjg@chromium.org>
At present when debugging is off, bootdev_find_in_blk() sometimes fails
to find a valid bootflow, e.g. with virtio. Accessing the 'blk' variable
later in the function seems to correct it.
Move the 'ret' check before the debug statement and set the block device
again aftewards, to work around this.
Signed-off-by: Simon Glass <sjg@chromium.org>
Enable some messages that might provide hints, but only for
CMD_BOOTFLOW_FULL since otherwise the -l flag is not available.
Signed-off-by: Simon Glass <sjg@chromium.org>
Suggested-by: Tom Rini <trini@konsulko.com>
Use the -l flag to indicate whether to report missing uclasses.
Also try to be more helpful when no devices are found. For example, when
we see something 'scsi0' requested and nothing was found, this indicates
that there are no SCSI devices, so show a suitable message.
Move messages out of the low-level functions so that silent operation
is possible.
This means they are never reported unless BOOTSTD_FULL is enabled, since
the -l flag cannot otherwise be set.
Suggested-by: Tom Rini <trini@konsulko.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
The current EFI implementation has a strange quirk where it watches
loaded files and uses the last-loaded file to determine the device that
is being booted from.
This is confusing with bootstd, where multiple options may exist. Even
loading a device tree will cause it to go wrong. There is no API for
passing this information, since the only entry into booting an EFI image
is the 'bootefi' command.
To work around this, call efi_set_bootdev() for EFI images, if possible,
just before booting.
Signed-off-by: Simon Glass <sjg@chromium.org>
Sometimes virtio may rely on PCI, or at least that is what the
distro_bootcmd script suggests. Add this in.
Signed-off-by: Simon Glass <sjg@chromium.org>
We should only store the FDT filename if we were able to determine one.
Adjust the logic for this.
This corrects the case where no FDT is needed to boot, such as with EFI
using ACPI.
Signed-off-by: Simon Glass <sjg@chromium.org>
- A few TI, mpc8379erdb and vexpress64 updates, some arm64 core updates,
test cleanups, Xen cleanups, assorted code and Kconfig cleanups (many
in preparation for using clang on ARM), psci pre-reloc fixes, enable
PCI bus mastering for ATA devices, and a gpio-uclass bugfix
If one leaves the CQSPI_REG_CMDCTRL in an unclean state this may cause
issues in future command reads. This issue came to light when some flash
reads in STIG mode were coming back dirty.
Co-developed-by: Apurva Nandan <a-nandan@ti.com>
Signed-off-by: Apurva Nandan <a-nandan@ti.com>
Signed-off-by: Dhruva Gole <d-gole@ti.com>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
OSPI controller supports all types of op variants in STIG mode,
only limitation being that the data payload should be less than
8 bytes when not using memory banks.
STIG mode is more stable for operations that send small data
payload and is more efficient than using DMA for few bytes of
memory accesses. It overcomes the limitation of minimum 4 bytes
read from flash into RAM seen in DAC mode.
Use STIG mode for all read and write operations that require
data input/output of less than 8 bytes from the flash, and thereby
support all four phases, cmd/address/dummy/data, through OSPI STIG.
Also, remove the reorder address chunk in apb_command_write since we now
setup ADDR BIT field that does the same job in a cleaner way.
Signed-off-by: Apurva Nandan <a-nandan@ti.com>
Signed-off-by: Dhruva Gole <d-gole@ti.com>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
buswidth and dtr fields in spi_mem_op are only valid when the
corresponding spi_mem_op phase has a non-zero length. For example,
SPI NAND core doesn't set buswidth when using SPI_MEM_OP_NO_ADDR
phase.
Fix the dtr checks in set_protocol() to ignore empty spi_mem_op
phases, as checking for dtr field in empty phase will result in
false negatives.
Signed-off-by: Apurva Nandan <a-nandan@ti.com>
Signed-off-by: Dhruva Gole <d-gole@ti.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
This patch adds fixups for s25fs512s to address the following issues
from reading SFDP:
- Non-uniform sectors by factory default. The setting needs to be
checked and assign erase hook as needed.
- Page size is wrongly advertised in SFDP.
- READ_1_1_2 (3Bh/3Ch), READ_1_1_4 (6Bh/6Ch), and PP_1_1_4 (32h/34h)
are not supported.
- Bank Address Register (BAR) is not supported.
In addition, volatile version of Quad Enable is used for safety.
Based on patch by Takahiro Kuwano with s25fs_s_post_bfpt_fixup() updated
to use 4-byte address commands instead of extended address mode and the
page_size is fixed to 256
For future use, manufacturer code should be moved out from framework
code as same as in Linux.
Reviewed-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Signed-off-by: Takahiro Kuwano <Takahiro.Kuwano@infineon.com>
Signed-off-by: Hai Pham <hai.pham.ud@renesas.com>
Signed-off-by: Cong Dang <cong.dang.xn@renesas.com>
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
By the time we jump to the err label, count represents the number of
gpios we've successfully requested. So by subtracting one, we fail to
free the most recently requested.
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Reviewed-by: Simon Glass <sjg@chromium.org>
Xen core_init() is calling HVC which should be called from EL1 level that's
why do Xen initialization only when U-Boot runs in EL1.
Signed-off-by: Michal Simek <michal.simek@amd.com>
U-Boot currently implements hypervisor calls only for ARM64 that's why add
dependency on ARM64. The hardware functionality is also available on
ARMv7a, but currently not supported by U-Boot, hence it is not added to the
list of dependencies here.
Signed-off-by: Michal Simek <michal.simek@amd.com>