When writing the serdes configuration register was moved to
mvneta_config_interface() the whole code block was removed from
mvneta_port_power_up() in the assumption that its only purpose was to
write the serdes configuration register. As mentioned by Russell King
its purpose was also to check for valid interface modes early so that
later in the driver we do not have to care for unexpected interface
modes.
Add back the test to let the driver bail out early on unhandled
interface modes.
Fixes: b4748553f5 ("net: ethernet: mvneta: Fix Serdes configuration for SoCs without comphy")
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Russell King <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
In mvneta_config_interface() the RGMII modes are catched by the default
case which is an error return. The RGMII modes are valid modes for the
driver, so instead of returning an error add a break statement to return
successfully.
This avoids this warning for non comphy SoCs which use RGMII, like
SolidRun Clearfog:
WARNING: CPU: 0 PID: 268 at drivers/net/ethernet/marvell/mvneta.c:3512 mvneta_start_dev+0x220/0x23c
Fixes: b4748553f5 ("net: ethernet: mvneta: Fix Serdes configuration for SoCs without comphy")
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Russell King <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
The older SoCs like Armada XP support a 2500BaseX mode in the datasheets
referred to as DR-SGMII (Double rated SGMII) or HS-SGMII (High Speed
SGMII). This is an upclocked 1000BaseX mode, thus
PHY_INTERFACE_MODE_2500BASEX is the appropriate mode define for it.
adding support for it merely means writing the correct magic value into
the MVNETA_SERDES_CFG register.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
The MVNETA_SERDES_CFG register is only available on older SoCs like the
Armada XP. On newer SoCs like the Armada 38x the fields are moved to
comphy. This patch moves the writes to this register next to the comphy
initialization, so that depending on the SoC either comphy or
MVNETA_SERDES_CFG is configured.
With this we no longer write to the MVNETA_SERDES_CFG on SoCs where it
doesn't exist.
Suggested-by: Russell King <[email protected]>
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
The Cypress cy15b104q and cy15v104q are 4Mbit serial SPI F-RAM devices.
Add support for them to the spi-nor driver.
The actual Device ID of this chip is 7f 7f 7f 7f 7f 7f c2 2c 04. That is
six times the continuation code 7f followed by c2 for Ramtron.
Unfortunately the chip sends the Device ID in reversed order, so the
continuation code is not at the beginning, but instead at the end. Even
more unfortunate is that when reading further the chip sends more 7f
codes which means we are not even able to count the continuation codes.
We can only hope that this reversed Device ID will never match any other
devices ID.
Collisions are improbable as of now, the solution from above is good
enough. In case of future collisions one can introduce an INFO9 macro,
with the downsize that struct flash_info would grow and we have lots of
flashes. A more elegant solution would be to introduce dedicated
flash ID tables for each bank in JESP106BA.
Signed-off-by: Sascha Hauer <[email protected]>
[[email protected]: amend commit description with possible
future solutions in case collisions occur.]
Signed-off-by: Tudor Ambarus <[email protected]>
The script addresses array increases with each new version. The driver
initializes the array to -EINVAL initially, but only up to the size
of the v1 array. Initialize the additional addresses for the newer
versions as well. Without this uninitialized values of the newer arrays
are treated as valid.
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Robin Gong <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>
The PWM iomem resource is optional and its presence indicates whether
the GPIO chip has a PWM or not, which is why mvebu_pwm_probe() returned
successfully when the PWM resource was not present. With f51b18d92b
the driver switched to devm_platform_ioremap_resource_byname() and
its error return is propagated to the caller, so now a missing PWM resource
leads to a probe error in the driver.
To fix this explicitly test for the presence of the PWM resource and
return successfully when it's not there. Do this check before the check
for the clock is done (which GPIO chips without a PWM do not have). Also
move the existing comment why the PWM resource is optional up to the
actual check.
Fixes: f51b18d92b ("gpio: mvebu: use devm_platform_ioremap_resource_byname()")
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Bartosz Golaszewski <[email protected]>
The jc42 driver passes I2C client's name as hwmon device name. In case
of device tree probed devices this ends up being part of the compatible
string, "jc-42.4-temp". This name contains hyphens and the hwmon core
doesn't like this:
jc42 2-0018: hwmon: 'jc-42.4-temp' is not a valid name attribute, please fix
This changes the name to "jc42" which doesn't have any illegal
characters.
Signed-off-by: Sascha Hauer <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Guenter Roeck <[email protected]>
In create_default_filesystem() when we allocate the idx node we must use
the idx_node_size we calculated just one line before, not tmp, which
contains completely other data.
Fixes: c4de6d7e43 ("ubifs: Refactor create_default_filesystem()")
Cc: [email protected] # v4.20+
Reported-by: Naga Sureshkumar Relli <[email protected]>
Tested-by: Naga Sureshkumar Relli <[email protected]>
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
set_seen() sets the bit corresponding to the PEB number in the bitmap,
so when self_check_seen() wants to find PEBs that haven't been seen we
have to print the PEBs that have their bit cleared, not the ones which
have it set.
Fixes: 5d71afb008 ("ubi: Use bitmaps in Fastmap self-check code")
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
The current descriptor is not on any list of the virtual DMA channel.
Once sdma_terminate_all() is called when a descriptor is currently
in flight then this one is forgotten to be freed. We have to call
vchan_terminate_vdesc() on this descriptor to re-add it to the lists.
Now that we also free the currently running descriptor we can (and
actually have to) remove the current descriptor from its list also
for the cyclic case.
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Robin Gong <[email protected]>
Tested-by: Robin Gong <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>
In sdma_tx_status() we must first find the current sdma_desc. In cyclic
mode we assume that this can always be found with vchan_find_desc().
This is true because do not remove the current descriptor from the
desc_issued list:
/*
* Do not delete the node in desc_issued list in cyclic mode, otherwise
* the desc allocated will never be freed in vchan_dma_desc_free_list
*/
if (!(sdmac->flags & IMX_DMA_SG_LOOP))
list_del(&vd->node);
We will change this in the next step, so check if the current descriptor is
the desired one also for the cyclic case.
Signed-off-by: Sascha Hauer <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>
vchan_vdesc_fini() shouldn't be called under a spin_lock. This is done
in two places, once in vchan_terminate_vdesc() and once in
vchan_synchronize(). Instead of freeing the vdesc right away, collect
the aborted vdescs on a separate list and free them along with the other
vdescs. The terminated descs are also freed in vchan_synchronize as done
before this patch.
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Peter Ujfalusi <[email protected]>
Tested-by: Peter Ujfalusi <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>
Originally freeing descriptors was split into a locked and an unlocked
part. The locked part in vchan_get_all_descriptors() collected all
descriptors on a separate list_head. This was done to allow iterating
over that new list in vchan_dma_desc_free_list() without a lock held.
This became broken in 13bb26ae88 ("dmaengine: virt-dma: don't always
free descriptor upon completion"). With this commit
vchan_dma_desc_free_list() no longer exclusively operates on the
separate list, but starts to put descriptors which can be reused back on
&vc->desc_allocated. This list operation should have been locked, but
wasn't.
In the mean time drivers started to call vchan_dma_desc_free_list() with
their lock held so that we now have the situation that
vchan_dma_desc_free_list() is called locked from some drivers and
unlocked from others.
To clean this up we have to do two things:
1. Add missing locking in vchan_dma_desc_free_list()
2. Make sure drivers call vchan_dma_desc_free_list() unlocked
This needs to be done atomically, so in this patch the locking is added
and all drivers are fixed.
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Peter Ujfalusi <[email protected]>
Tested-by: Peter Ujfalusi <[email protected]>
Reviewed-by: Green Wan <[email protected]>
Tested-by: Green Wan <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Vinod Koul <[email protected]>
ata_qc_complete_multiple() is called with a mask of the still active
tags.
mv_sata doesn't have this information directly and instead calculates
the still active tags from the started tags (ap->qc_active) and the
finished tags as (ap->qc_active ^ done_mask)
Since 28361c4036 the hw_tag and tag are no longer the same and the
equation is no longer valid. In ata_exec_internal_sg() ap->qc_active is
initialized as 1ULL << ATA_TAG_INTERNAL, but in hardware tag 0 is
started and this will be in done_mask on completion. ap->qc_active ^
done_mask becomes 0x100000000 ^ 0x1 = 0x100000001 and thus tag 0 used as
the internal tag will never be reported as completed.
This is fixed by introducing ata_qc_get_active() which returns the
active hardware tags and calling it where appropriate.
This is tested on mv_sata, but sata_fsl and sata_nv suffer from the same
problem. There is another case in sata_nv that most likely needs fixing
as well, but this looks a little different, so I wasn't confident enough
to change that.
Fixes: 28361c4036 ("libata: add extra internal command")
Cc: [email protected]
Tested-by: Pali Rohár <[email protected]>
Signed-off-by: Sascha Hauer <[email protected]>
Add missing export of ata_qc_get_active(), as per Pali.
Signed-off-by: Jens Axboe <[email protected]>
When a new fastmap is about to be written UBI must make sure it has a
free block for a fastmap anchor available. For this ubi_update_fastmap()
calls ubi_ensure_anchor_pebs(). This stopped working with 2e8f08deab
("ubi: Fix races around ubi_refill_pools()"), with this commit the wear
leveling code is blocked and can no longer produce free PEBs. UBI then
more often than not falls back to write the new fastmap anchor to the
same block it was already on which means the same erase block gets
erased during each fastmap write and wears out quite fast.
As the locking prevents us from producing the anchor PEB when we
actually need it, this patch changes the strategy for creating the
anchor PEB. We no longer create it on demand right before we want to
write a fastmap, but instead we create an anchor PEB right after we have
written a fastmap. This gives us enough time to produce a new anchor PEB
before it is needed. To make sure we have an anchor PEB for the very
first fastmap write we call ubi_ensure_anchor_pebs() during
initialisation as well.
Fixes: 2e8f08deab ("ubi: Fix races around ubi_refill_pools()")
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
Since ceeeb99cd8 we no longer abuse the DMA_CTRL_ACK flag for custom
driver use and introduced the MXS_DMA_CTRL_WAIT4END instead. We have not
changed all users to this flag though. This patch fixes it for the
mxs-mmc driver.
Fixes: ceeeb99cd8 ("dmaengine: mxs: rename custom flag")
Signed-off-by: Sascha Hauer <[email protected]>
Tested-by: Fabio Estevam <[email protected]>
Reported-by: Bruno Thomsen <[email protected]>
Tested-by: Bruno Thomsen <[email protected]>
Cc: [email protected]
Signed-off-by: Ulf Hansson <[email protected]>
omap_elm_correct_data() returns the number of bitflips for the whole
page. This is wrong, it should return the maximum number of bitflips
found in each ECC step.
In my case with a 4k page size NAND mtcdore reported -EUCLEAN with
only 12 bitflips on a page where we could correct up to 128 bits per
page (provided they are distributed equally on the 8 ECC steps)
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Miquel Raynal <[email protected]>
Switch from module_platform_driver_probe() to module_platform_driver().
The former is not suitable for booting with device tree as the driver
will be registered before the device and thus won't be probed again
when the device is present.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Shawn Guo <[email protected]>
integrity_kernel_read() can fail in which case we forward to call
ahash_request_free() on a currently running request. We have to wait
for its completion before we can free the request.
This was observed by interrupting a "find / -type f -xdev -print0 | xargs -0
cat 1>/dev/null" with ctrl-c on an IMA enabled filesystem.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Mimi Zohar <[email protected]>
integrity_kernel_read() returns the number of bytes read. If this is
a short read then this positive value is returned from
ima_calc_file_hash_atfm(). Currently this is only indirectly called from
ima_calc_file_hash() and this function only tests for the return value
being zero or nonzero and also doesn't forward the return value.
Nevertheless there's no point in returning a positive value as an error,
so translate a short read into -EINVAL.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Mimi Zohar <[email protected]>
HMACs can only be generated on the system the UBIFS image is running on.
To support offline signed images we add a PKCS#7 signature to the UBIFS
image which can be created by mkfs.ubifs.
Both the master node and the superblock need to be authenticated, during
normal runtime both are protected with HMACs. For offline signature
support however only a single signature is desired. We add a signature
covering the superblock node directly behind it. To protect the master
node a hash of the master node is added to the superblock which is used
when the master node doesn't contain a HMAC.
Transition to a read/write filesystem is also supported. During
transition first the master node is rewritten with a HMAC (implicitly,
it is written anyway as the FS is marked dirty). Afterwards the
superblock is rewritten with a HMAC. Once after the image has been
mounted read/write it is HMAC only, the signature is no longer required
or even present on the filesystem.
In an offline signed image the master node is authenticated by the
superblock. In a transition to r/w we have to make sure that the master
node is rewritten before the superblock node. In this case the master
node gets a HMAC and its authenticity no longer depends on the
superblock node. There are some cases in which the current code first
writes the superblock node though, so with this patch writing of the
superblock node is delayed until the master node is written.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
This adds serdev support to the Marvell hci uart driver. Only basic
serdev support, none of the fancier features like regulator or enable
GPIO support is added for now.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Marcel Holtmann <[email protected]>
For the Marvell HCI UART we have to upload two firmware files. The first
one is only for switching the baudrate of the device to a higher
baudrate. After the baudrate switching firmware has been uploaded the
device waits for a final ack (0x5a) before actually switching the
baudrate. To send this final ack with the old baudrate give the hci
ldisc workqueue a chance to run before switching the baudrate. Without
this the final ack will never be received by the device and firmware
upload fails.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Marcel Holtmann <[email protected]>
The hci UART line discipline sends its characters in a workqueue. Some
devices like the Marvell Bluetooth chips need to make sure that all
queued characters are sent before switching the baudrate. This adds
a function to synchronize with the workqueue.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Marcel Holtmann <[email protected]>
The gpmi driver performance suffers from NAND operations being split
in multiple small DMA transfers. This has been forced by the NAND layer
in the former days, but now with exec_op we can use the controller as
intended.
With this patch gpmi_nfc_exec_op becomes the main entry point to NAND
operations. Here all instructions are collected and chained as separate
DMA transfers. In the end whole chain is fired and waited to be
finished. gpmi_nfc_exec_op only does the hardware operations, bad block
marker swapping and buffer scrambling is done by the callers. It's worth
noting that the nand_*_op functions always take the buffer lengths for
the data that the NAND chip actually transfers. When doing BCH we have
to calculate the net data size from the raw data size in some places.
This patch has been tested with 2048/64 and 2048/128 byte NAND on
i.MX6q. mtd_oobtest, mtd_subpagetest and mtd_speedtest run without
errors. nandbiterrs, nandpagetest and nandsubpagetest userspace tests
from mtdutils run without errors and UBIFS can successfully be mounted.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Miquel Raynal <[email protected]>
The mxs dma driver uses the flags parameter in dmaengine_prep_slave_sg() for
custom flags, but still uses the dmaengine specific names of the flags.
Do a little bit better and at least give the flag a custom name.
Signed-off-by: Sascha Hauer <[email protected]>
Acked-by: Vinod Koul <[email protected]>
Signed-off-by: Miquel Raynal <[email protected]>
The mxs dma driver can do PIO transfers. A pointer to the PIO words
to transfer is passed in the struct scatterlist * argument of
dmaengine_prep_slave_sg(). It's quite ugly and non obvious to cast
u32 * to struct scatterlist * each time when calling
dmaengine_prep_slave_sg(), so add a static inline wrapper function
to be called by the user along with a description what is going on.
Signed-off-by: Sascha Hauer <[email protected]>
Acked-by: Vinod Koul <[email protected]>
Signed-off-by: Miquel Raynal <[email protected]>
The DMA_PREP_INTERRUPT flag is no longer needed by the mxs DMA driver,
drop it.
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Miquel Raynal <[email protected]>
Signed-off-by: Miquel Raynal <[email protected]>
The mxs dma driver insists on having the DMA_PREP_INTERRUPT flag set
on all but the first transfer. There's no need to let the user set this
flag, the driver can do it internally whenever it needs it. Drop
handling of this flag from the driver.
Signed-off-by: Sascha Hauer <[email protected]>
Acked-by: Vinod Koul <[email protected]>
Signed-off-by: Miquel Raynal <[email protected]>
The gpmi driver aggressively en/disables the clocks between operations
which has significant performance cost. Use runtime PM to get rid of
this bottleneck.
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Miquel Raynal <[email protected]>
Signed-off-by: Miquel Raynal <[email protected]>
The i.MX23 specific option read code is called right after nand_scan. We
can rely on the NAND core having disabled the chipselect, so there's no
point in restoring the original chip select after NAND operations. Drop
it.
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Miquel Raynal <[email protected]>
Signed-off-by: Miquel Raynal <[email protected]>
gpmi_ecc_read_page_data uses the page parameter only for a debug printf,
so we can drop the parameter and the debug printf. Moving the oob
delivery from gpmi_ecc_read_page_data to gpmi_ecc_read_page makes the
oob_required parameter unnecessary aswell.
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Miquel Raynal <[email protected]>
Signed-off-by: Miquel Raynal <[email protected]>
The driver calls nand_read_page_op without a buffer passed and then
calls chip->legacy.read_buf to read the buffer afterwards which is
the same as passing the buffer nand_read_page_op in the first place.
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Miquel Raynal <[email protected]>
Signed-off-by: Miquel Raynal <[email protected]>
this->page_buffer_virt and this->payload_virt are always set to the same
value, so drop the former and just use the latter. Same for
this->page_buffer_virt and this->payload_virt.
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Miquel Raynal <[email protected]>
Signed-off-by: Miquel Raynal <[email protected]>
This moves the whole driver into a single C file. The filename gpmi-lib
implies that it implements library functions, but in fact there are
several cases where functions in gpmi-lib.c call back into functions in
gpmi-nand.c. With this one has to constantly jump between those two
files, so moving it into a single file improves readability, even when
the file gets quite large.
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Miquel Raynal <[email protected]>
Signed-off-by: Miquel Raynal <[email protected]>
The NAND core has a NAND operation tracing function, but it can only
be used by drivers using the generic option parser from the NAND core.
Export the tracing function as a static inline function in rawnand.h
so that drivers implementing exec_op directly do not have to write their
own operation tracing.
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Miquel Raynal <[email protected]>
Signed-off-by: Miquel Raynal <[email protected]>
The Kernel has nice hexdump facilities, use them rather a homebrew
hexdump function.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Tyler Hicks <[email protected]>
caam_dump_sg() is only compiled in when DEBUG is defined, hence the
messages are debug messages. Remove the @level argument from
caam_dump_sg() and print all messages at debug level.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
The CAAM driver used to put its debug messages inside #ifdef DEBUG and
then prints the messages at KERN_ERR level. Replace this with proper
functions printing at KERN_DEBUG level. The #ifdef DEBUG gets
unnecessary when the right functions are used.
This replaces:
- print_hex_dump(KERN_ERR ...) inside #ifdef DEBUG with
print_hex_dump_debug(...)
- dev_err() inside #ifdef DEBUG with dev_dbg()
- printk(KERN_ERR ...) inside #ifdef DEBUG with dev_dbg()
Some parts of the driver use these functions already, so it is only
consequent to use the debug function consistently.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
The CAAM driver defines its own debug() macro, but it is unused. Remove
it.
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Horia Geantă <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
Since eaed71a44a ("crypto: caam - add ecb(*) support") the IV can be
NULL, so only dump it when it's non NULL as designated by the ivsize
variable.
Fixes: eaed71a44a ("crypto: caam - add ecb(*) support")
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
For encryption the destination pointer was still mapped, so the hex dump
may be wrong. The IV still contained the input IV while printing instead
of the output IV as intended.
For decryption the destination pointer was still mapped, so the hex dump
may be wrong. The IV dump was correct.
Do the hex dumps consistenly after the buffers have been unmapped and
in case of IV copied to their final destination.
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Horia Geantă <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
Each time the DMA engine signals a transaction error the driver prints
a message at error level. Getting transaction errors is pretty much
expected on baudrate mismatches and the correspoding error counters
are increased in this case properly. Remove the error message which
is possibly repeated at a very high rate which can lock up the whole
system.
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Fabio Estevam <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>
memorg->ntargets is initialized with '1'. It should be initialized with
the maxchips argument from nand_scan() instead. Otherwise multi chip
support errors out on the secondary chip selects when trying to call
nand_reset() on them:
WARNING: CPU: 0 PID: 1 at drivers/mtd/nand/raw/internals.h:114
nand_reset_op+0x194/0x1c4
With this memorg->ntargets is initialized with the maximum number of
chip selects supported by the driver. After having detected the number
of actually connected chips memory->ntargets is updated with that
number.
Fixes: 32813e2884 ("mtd: rawnand: Get rid of chip->numchips")
Signed-off-by: Sascha Hauer <[email protected]>
Reviewed-by: Boris Brezillon <[email protected]>
Signed-off-by: Miquel Raynal <[email protected]>
in dbg_walk_index ubifs_load_znode is used to load the znode behind
a zbranch. ubifs_load_znode links the new child znode to the zbranch,
so doing it again is unnecessary.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
ifdefs reduce readability and compile coverage. This removes the ifdefs
around CONFIG_UBIFS_ATIME_SUPPORT by replacing them with IS_ENABLED()
where applicable. The fs layer would fall back to generic_update_time()
when .update_time doesn't exist. We do this fallback explicitly now.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
ifdefs reduce readablity and compile coverage. This removes the ifdefs
around CONFIG_FS_ENCRYPTION by using IS_ENABLED and relying on static
inline wrappers. A new static inline wrapper for setting sb->s_cop is
introduced to allow filesystems to unconditionally compile in their
s_cop operations.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
UBIFS bails out early from try_read_node() when it doesn't have to check
the CRC. Still the node hash has to be checked, otherwise wrong data
could be sneaked into the FS. Fix this by not bailing out early and
always checking the node hash.
Fixes: 16a26b20d2 ("ubifs: authentication: Add hashes to index nodes")
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
With the preparations all being done this patch now enables authentication
support for UBIFS. Authentication is enabled when the newly introduced
auth_key and auth_hash_name mount options are passed. auth_key provides
the key which is used for authentication whereas auth_hash_name provides
the hashing algorithm used for this FS. Passing these options make
authentication mandatory and only UBIFS images that can be authenticated
with the given key are allowed.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
In authenticated mode we cannot fixup the inode sizes in-place
during recovery as this would invalidate the hashes and HMACs
we stored for this inode.
Instead, we just write the updated inodes to the journal. We can
only do this after ubifs_rcvry_gc_commit() is done though, so for
authenticated mode call ubifs_recover_size() after
ubifs_rcvry_gc_commit() and not vice versa as normally done.
Calling ubifs_recover_size() after ubifs_rcvry_gc_commit() has the
drawback that after a commit the size fixup information is gone, so
when a powercut happens while recovering from another powercut
we may lose some data written right before the first powercut.
This is why we only do this in authenticated mode and leave the
behaviour for unauthenticated mode untouched.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
This patch calculates the necessary hashes and HMACs for the default
filesystem so that the dynamically created default fs can be
authenticated.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
This adds a HMAC covering the super block node and adds the logic that
decides if a filesystem shall be mounted unauthenticated or
authenticated.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
During creation of the default filesystem on an empty flash the default
LPT is created. With this patch a hash over the default LPT is
calculated which can be added to the default filesystems master node.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
The master node contains hashes over the root index node and the LPT.
This patch adds a HMAC to authenticate the master node itself.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
The LPT needs to be authenticated aswell. Since the LPT is only written
during commit it is enough to authenticate the whole LPT with a single
hash which is stored in the master node. Only the leaf nodes (pnodes)
are hashed which makes the implementation much simpler than it would be
to hash the complete LPT.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
Make sure that during replay all buds can be authenticated. To do
this we calculate the hash chain until we find an authentication
node and check the HMAC in that node against the current status
of the hash chain.
After a power cut it can happen that some nodes have been written, but
not yet the authentication node for them. These nodes have to be
discarded during replay.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
To be able to authenticate the garbage collector journal head add
authentication nodes to the buds the garbage collector creates.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
Nodes that are written to flash can only be authenticated through the
index after the next commit. When a journal replay is necessary the
nodes are not yet referenced by the index and thus can't be
authenticated.
This patch overcomes this situation by creating a hash over all nodes
beginning from the commit start node over the reference node(s) and
the buds themselves. From
time to time we insert authentication nodes. Authentication nodes
contain a HMAC from the current hash state, so that they can be
used to authenticate a journal replay up to the point where the
authentication node is. The hash is continued afterwards
so that theoretically we would only have to check the HMAC of
the last authentication node we find.
Overall we get this picture:
,,,,,,,,
,......,...........................................
,. CS , hash1.----. hash2.----.
,. | , . |hmac . |hmac
,. v , . v . v
,.REF#0,-> bud -> bud -> bud.-> auth -> bud -> bud.-> auth ...
,..|...,...........................................
, | ,
, | ,,,,,,,,,,,,,,,
. | hash3,----.
, | , |hmac
, v , v
, REF#1 -> bud -> bud,-> auth ...
,,,|,,,,,,,,,,,,,,,,,,
v
REF#2 -> ...
|
V
...
Note how hash3 covers CS, REF#0 and REF#1 so that it is not possible to
exchange or skip any reference nodes. Unlike the picture suggests the
auth nodes themselves are not hashed.
With this it is possible for an offline attacker to cut each journal
head or to drop the last reference node(s), but not to skip any journal
heads or to reorder any operations.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
With this patch the hashes over the index nodes stored in the tree node
cache are written to flash and are checked when read back from flash.
The hash of the root index node is stored in the master node.
During journal replay the hashes are regenerated from the read nodes
and stored in the tree node cache. This means the nodes must previously
be authenticated by other means. This is done in a later patch.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
As part of the UBIFS authentication support every branch in the index
gets a hash covering the referenced node. To make that happen the tree
node cache needs hashes over the nodes. This patch adds a hash argument
to ubifs_tnc_add() and ubifs_tnc_add_nm(). The hashes are calculated
from the callers of these functions which actually prepare the nodes.
With this patch all the leaf nodes of the index tree get hashes, but
currently nothing is done with these hashes, this is left for a later
patch.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
With authentication support some nodes (master node, super block node)
get a HMAC embedded into them. This patch adds functions to prepare and
write such a node.
The difficulty is that besides the HMAC the nodes also have a CRC which
must stay valid. This means we first have to initialize all fields in
the node, then calculate the HMAC (not covering the CRC) and finally
calculate the CRC.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
This patch adds the various helper functions needed for authentication
support. We need functions to hash nodes, to embed HMACs into a node and
to compare hashes and HMACs. Most functions first check if this
filesystem is authenticated and bail out early if not, which makes the
functions safe to be called with disabled authentication.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
When adding authentication support we will embed a HMAC into some
nodes. To prepare these nodes we have to first initialize the nodes,
then add a HMAC and finally add a CRC. To accomplish this add separate
ubifs_init_node/ubifs_crc_node functions.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
This patch adds the changes to the on disk format needed for
authentication support. We'll add:
* a HMAC covering super block node
* a HMAC covering the master node
* a hash over the root index node to the master node
* a hash over the LPT to the master node
* a flag to the filesystem flag indicating the filesystem is
authenticated
* an authentication node necessary to authenticate the nodes written
to the journal heads while they are written.
* a HMAC of a well known message to the super block node to be able
to check if the correct key is provided
And finally, not visible in this patch, nevertheless explained here:
* hashes over the referenced child nodes in each branch of a index node
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
The superblock node is read/modified/written several times throughout
the UBIFS code. Instead of reading it from the device each time just
keep a copy in memory and write back the modified copy when necessary.
This patch helps for authentication support, here we not only have to
read the superblock node, but also have to authenticate it, which
is easier if we do it once during initialization.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
write_node() is used only once and can easily be replaced with calls
to ubifs_prepare_node()/write_head() which makes the code a bit shorter.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
ubifs_lpt_lookup() starts by looking up the nth pnode in the LPT. We
already have this functionality in ubifs_pnode_lookup(). Use this
function rather than open coding its functionality.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
ubifs_lpt_lookup could be implemented using pnode_lookup. To make that
possible move pnode_lookup from lpt.c to lpt_commit.c. Rename it to
ubifs_pnode_lookup since it's now exported.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
read_znode() takes len, lnum and offs arguments which the caller all
extracts from the same struct ubifs_zbranch *. When adding authentication
support we would have to add a pointer to a hash to the arguments which
is also part of struct ubifs_zbranch. Pass the ubifs_zbranch * instead
so that we do not have to add another argument.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
try_read_node() takes len, lnum and offs arguments which the caller all
extracts from the same struct ubifs_zbranch *. When adding authentication
support we would have to add a pointer to a hash to the arguments which
is also part of struct ubifs_zbranch. Pass the ubifs_zbranch * instead
so that we do not have to add another argument.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
create_default_filesystem() allocates memory for a node, writes that
node and frees the memory directly afterwards. With this patch we
allocate memory for all nodes at the beginning of the function and
free the memory at the end. This makes it easier to implement
authentication support since with authentication support we'll need
the contents of some nodes when creating other nodes.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
The following sequence triggers
ubifs_assert(c, c->lst.taken_empty_lebs > 0);
at the end of ubifs_remount_fs():
mount -t ubifs /dev/ubi0_0 /mnt
echo 1 > /sys/kernel/debug/ubifs/ubi0_0/ro_error
umount /mnt
mount -t ubifs -o ro /dev/ubix_y /mnt
mount -o remount,ro /mnt
The resulting
UBIFS assert failed in ubifs_remount_fs at 1878 (pid 161)
is a false positive. In the case above c->lst.taken_empty_lebs has
never been changed from its initial zero value. This will only happen
when the deferred recovery is done.
Fix this by doing the assertion only when recovery has been done
already.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
The maximum CPU frequency for the i.MX53 QSB is 1GHz, so disable the
1.2GHz OPP. This makes the board work again with configs that have
cpufreq enabled like imx_v6_v7_defconfig on which the board stopped
working with the addition of cpufreq-dt support.
Fixes: 791f416608 ("ARM: dts: imx53: add cpufreq-dt support")
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Shawn Guo <[email protected]>
This is a preparation step to make the adding of virt-dma easier.
We create a struct sdma_desc, move some fields from struct sdma_channel
there and add a pointer from the former to the latter. For now we
allocate the data statically in struct sdma_channel, but with
virt-dma support it will be dynamically allocated.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Robin Gong <[email protected]>
Reviewed-by: Sascha Hauer <[email protected]>
Tested-by: Lucas Stach <[email protected]>
Signed-off-by: Vinod Koul <[email protected]>
The comment above pnode_lookup claims the range for the pnode number is
from 0 to main_lebs - 1. This is wrong because every pnode has
informations about UBIFS_LPT_FANOUT LEBs, thus the corrent range is
0 to to (main_lebs - 1) / UBIFS_LPT_FANOUT.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
"point of view" makes more sense than "point of few". Fix this.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
In the description of reserve_space() it is claimed that write_node()
and write_head() unlock the journal head. This is not true and has never
been true. All callers of write_node() and write_head() call
release_head() themselves. Remove the wrong comment.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Richard Weinberger <[email protected]>
Use this->auxiliary_virt and this->auxiliary_phys directly rather
than creating extra local variables for them.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <[email protected]>
Signed-off-by: Boris Brezillon <[email protected]>
The caller of bch_set_geometry() expects the return value to
be an error code, so !0 is not valid. return the error from the
just called function instead.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <[email protected]>
Signed-off-by: Boris Brezillon <[email protected]>
Instead of putting direct_dma_map_ok into driver struct pass it around
between functions to make the code more readable.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <[email protected]>
Signed-off-by: Boris Brezillon <[email protected]>
read_page_prepare(), read_page_end() and read_page_swap_end() are
trivial functions that are used only once and take 8 arguments each.
De-obfuscate the code by open coding these functions in
gpmi_ecc_read_page()
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <[email protected]>
Signed-off-by: Boris Brezillon <[email protected]>
Instead of putting the buffer and len passed in from the mtd core
into the private data struct, just pass it around in the GPMI
drivers functions. This makes the lifetime of the variables more
clear and the code easier to follow.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <[email protected]>
Signed-off-by: Boris Brezillon <[email protected]>
The GPMI nand driver puts dma_ops_type in its private data struct. Based
on the ops type the DMA callback handler unmaps previously mapped
buffers. Instead of unmapping the buffers in the DMA callback handler,
do this in the caller directly which waits for the DMA transfer to
finish. This makes the whole dma_ops_type mechanism unnecessary.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <[email protected]>
Signed-off-by: Boris Brezillon <[email protected]>
EVM might update the evm xattr while the VFS performs a remount to
readonly mode. This is not properly checked for, additionally check
the s_readonly_remount superblock flag before writing.
The bug can for example be observed with UBIFS. UBIFS checks the free
space on the device before and after a remount. With EVM enabled the
free space sometimes differs between both checks.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Mimi Zohar <[email protected]>
Since we now have our own read_page/write_page functions
mxc_nand_command() will no longer be called with NAND_CMD_READ0,
NAND_CMD_READOOB, NAND_CMD_SEQIN and NAND_CMD_PAGEPROG. Remove
the code handling these commands.
Signed-off-by: Sascha Hauer <[email protected]>
Signed-off-by: Boris Brezillon <[email protected]>
Sascha Hauer