| Age | Commit message (Collapse) | Author |
|---|
|
Some chips like Exynos4412 have fixed internal CIU clock divider.
Instead of reading it from non-standard "div" dts property, store its
value in the driver internally, in static chip data associated with
corresponding compatible. This makes it possible to avoid using
host->div for storing it, so the latter can be removed safely. Also
create a helper function called exynos_dwmmc_get_ciu_div() for getting
the current div value: in case the fixed div is provided in the chip
data it will be used, otherwise the current div value is being read from
CLKSEL register.
The insights for this change were taken from dw_mmc-exynos.c driver in
Linux kernel.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
CLKSEL register offset may vary between different Exynos chips, e.g. on
ARM64 vs ARM32 chips. Provide a way to specify its offset value for each
compatible instead of hard-coding its value in read/write calls.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
The obsolete "samsung,timing" dts property is now split into
"samsung,dw-mshc-ciu-div" (for holding the internal DW MMC divider
value) and "samsung,dw-mshc-sdr-timing" (for actual timing values) in
upstream Linux kernel. Rework the driver to make use of new properties
instead of the old one. All affected dts files were already updated
accordingly.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Update the driver to use livetree API instead of FDT one.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
exynos_dwmci_get_config() is called from the probe function and used to
read data from device tree. Make use of .of_to_plat driver callback
instead, and convert exynos_dwmci_get_config() to match its signature.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
New Exynos chips should implement clock drivers using CCF framework. In
that case corresponding CCF functions can be used to get/set the clock
rates. Moreover, already existing get_mmc_clk() and set_mmc_clk() calls
are only implemented for CONFIG_CPU_V7A (i.e. ARM32 chips). In case of
ARM64 chips that config option is not defined, so build will crash on
linking stage, with errors like these:
ld: drivers/mmc/exynos_dw_mmc.o:
in function `exynos_dwmci_get_sclk':
undefined reference to `get_mmc_clk'
ld: drivers/mmc/exynos_dw_mmc.o:
in function `exynos_dwmci_set_sclk':
undefined reference to `set_mmc_clk'
Fix that issue by using CCF clocks API on ARM64 platforms for getting
and setting the source clock (sclk = SDCLKIN = CIU) rate. To implement
this, first extract the existing ARM32 clock control code into helper
functions with more generic signatures to abstract getting/setting the
sclk rate. Then add CCF clock support to those functions for ARM64
platforms.
Fixes: a082a2dde061 ("EXYNOS5: DWMMC: Added FDT support for DWMMC")
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Pinmux configuration on ARM64 platforms must be performed during startup
in pinctrl driver using info from device tree. exynos_pinmux_config()
and pinmux_decode_periph_id() are only available on ARM32 platforms, so
don't call those functions on ARM64 platforms. Instead of the latter
function, use "non-removable" property from device tree to derive the
dev_index value.
This fixes next linking errors on ARM64 platforms:
ld: drivers/mmc/exynos_dw_mmc.o:
in function `exynos_dwmci_get_config':
undefined reference to `pinmux_decode_periph_id'
ld: drivers/mmc/exynos_dw_mmc.o:
in function `do_dwmci_init':
undefined reference to `exynos_pinmux_config'
Fixes: a082a2dde061 ("EXYNOS5: DWMMC: Added FDT support for DWMMC")
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
In case of CONFIG_DM_MMC, host->priv actually holds (struct udevice *),
and not (struct dwmci_exynos_priv_data *). This makes *priv pointer
invalid and may lead to Synchronous Abort during its dereference later
in exynos_dwmci_board_init(). Fix it by extracting
exynos_dwmmc_get_priv() helper from exynos_dwmci_clksel() and using it
for getting the private data in exynos_dwmci_board_init()
Fixes: 3537ee879e04 ("mmc: exynos_dw_mmc: support the Driver mode for Exynos")
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Getting the base address with outdated fdtdec_get_addr() API and further
casting it to (void *) leads to next build warning on ARM64 platforms:
In function 'exynos_dwmci_get_config':
warning: cast to pointer from integer of different size
[-Wint-to-pointer-cast]
host->ioaddr = (void *)base;
Use livetree API instead (dev_read_addr_ptr()), which handles this
correctly.
Fixes: a082a2dde061 ("EXYNOS5: DWMMC: Added FDT support for DWMMC")
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Add missing header guard to prevent possible build errors.
Fixes: 77b55e8cfcee ("ARM: exynos: move SoC sources to mach-exynos")
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Update the bindings doc for Exynos DW MMC block to follow the upstream
example and reflect the latest changes made in corresponding Linux
kernel bindings.
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Some device tree properties for DW MMC block were updated in Linux
kernel. Let's follow its example and rework corresponding properties in
all Exynos device trees. Don't remove outdated properties yet, it'll be
done later once DW MMC driver is updated accordingly to read the updated
properties instead of outdated ones.
Next properties are added:
* samsung,dw-mshc-ciu-div and samsung,dw-mshc-sdr-timing:
They were derived from outdated samsung,timing property.
* fifo-depth (generic replacement for fifoth_val):
FIFO depth was calculated from fifoth_val (using expressions from
FIFOTH register description in TRM):
fifo-depth = ((fifoth_val >> 16) + 1) * 2
* bus-width: generic replacement for samsung,bus-width
* clock-frequency: generic replacement for bus_hz
* non-removable: generic replacement for samsung,removable = <0>
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Fix most of checkpatch warnings and other obvious style issues.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Reviewed-by: Quentin Schulz <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Rework kernel-doc comments in dwmmc.h header so it's actually possible
to generate a proper documentation from it usin scripts/kernel-doc
script, with no errors.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Replace fifoth_val property with its fifo-depth counterpart in all DW
MMC drivers. fifo-depth is a common property used in upstream Linux
kernel. The FIFOTH register value will be calculated using fifo-depth
value in DW MMC core (dw_mmc.c). This change reduces code duplication in
platform drivers, and pulls common FIFOTH register value calculation
into core dw_mmc driver where it belongs.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Some DW MMC blocks (e.g. those on modern Exynos chips) support 64-bit
DMA addressing mode. 64-bit DW MMC variants differ from their 32-bit
counterparts:
- the register layout is a bit different (because there are additional
IDMAC registers present for storing upper part of 64-bit addresses)
- DMA descriptor structure is bigger and different from 32-bit one
Introduce all necessary changes to enable support for 64-bit DMA capable
DW MMC blocks. Next changes were made:
1. Check which DMA address mode is supported in current IP-core
version. HCON register (bit 27) indicates whether it's 32-bit or
64-bit addressing. Add boolean .dma_64bit_address field to struct
dwmci_host and store the result there. dwmci_init_dma() function is
introduced for doing so, which is called on driver's init.
2. Add 64-bit DMA descriptor (struct dwmci_idmac64) and use it in
dwmci_prepare_desc() in case if .dma_64bit_address field is true.
A new dwmci_set_idma_desc64() function was added for populating that
descriptor.
3. Add registers for 64-bit DMA capable blocks. To make the access to
IDMAC registers universal between 32-bit / 64-bit cases, a new
struct dwmci_idmac_regs (and corresponding host->regs field) was
introduced, which abstracts the hardware by being set to
appropriate offset constants on init. All direct calls to IDMAC
registers were correspondingly replaced by accessing host->regs.
4. Allocate and use 64-bit DMA descriptors buffer in case when IDMAC
is 64-bit capable. Extract all the code (except for the IDMAC
descriptors buffer allocation) from dwmci_send_cmd() to
dwmci_send_cmd_common(), so that it's possible to keep IDMAC
buffer (either 32-bit or 64-bit) on stack during send_cmd routine.
The insights for this implementation were taken from Linux kernel DW MMC
driver.
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Prepare for adding 64-bit IDMAC descriptors by renaming current 32-bit
descriptor and its fields accordingly. While at it, make use of
virt_to_phys() to make it more obvious in which places the physical
addresses have to be used.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Make dwmci_prepare_data() function easier to read by extracting the
preparation of IDMAC descriptor into a dedicated function.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Reviewed-by: Quentin Schulz <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Make dwmci_send_cmd() easier to read by moving the DMA transfer handling
code into a dedicated function.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Reviewed-by: Quentin Schulz <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
FIFO data transfer is implemented as quite a massive chunk of code.
Extract it into a dedicated function to make dwmci_data_transfer()
easier to read and reduce the indentation level of the code.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Extract the clock divider update into dwmci_update_div() function. It's
a procedure recommended in TRM, so it's better to keep it in a dedicated
function to make the code clearer.
While at it also extract the clock control code into a separate routine
to avoid code duplication in dwmci_setup_bus().
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Move FIFO threshold initialization into a separate function to make
dwmci_init() more readable.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Reviewed-by: Quentin Schulz <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Waiting for data busy is a logically separate operation and should be
implemented as a separate routine. Follow Linux kernel example and
extract it from dwmci_send_cmd(). This way it doesn't clutter
dwmci_send_cmd() function, and can be reused later in other cases.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Reviewed-by: Quentin Schulz <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
struct idmac is only used in dw_mmc.c, so move it there from dwmmc.h to
avoid cluttering the interface in the header.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Reviewed-by: Quentin Schulz <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Nobody seems to use it, so just remove it.
No functional change.
Signed-off-by: Sam Protsenko <[email protected]>
Reviewed-by: Quentin Schulz <[email protected]>
Signed-off-by: Minkyu Kang <[email protected]>
|
|
Currently, zalloc() calls uncondtionally memset(),
if the allocation failes, memset() will write to a null pointer.
Fix by using kzalloc().
Signed-off-by: Richard Weinberger <[email protected]>
Reviewed-by: Heinrich Schuchardt <[email protected]>
|
|
Both gcc and clang support this for a long time.
Make sure the feature is present.
Signed-off-by: Richard Weinberger <[email protected]>
|
|
While zalloc() takes a size_t type, adding 1 to the le32 variable
will overflow.
A carefully crafted ext4 filesystem can exhibit an inode size of 0xffffffff
and as consequence zalloc() will do a zero allocation.
Later in the function the inode size is again used for copying data.
So an attacker can overwrite memory.
Avoid the overflow by using the __builtin_add_overflow() helper.
Signed-off-by: Richard Weinberger <[email protected]>
|
|
res needs to be large enough to store both strings rem and target,
plus the path separator and the terminator.
Currently the space for the path separator is not accounted, so
the heap is corrupted by one byte.
Signed-off-by: Richard Weinberger <[email protected]>
Reviewed-by: Miquel Raynal <[email protected]>
|
|
The squashfs driver blindly follows symlinks, and calls sqfs_size()
recursively. So an attacker can create a crafted filesystem and with
a deep enough nesting level a stack overflow can be achieved.
Fix by limiting the nesting level to 8.
Signed-off-by: Richard Weinberger <[email protected]>
Reviewed-by: Miquel Raynal <[email protected]>
|
|
The function can fail and return NULL.
Signed-off-by: Richard Weinberger <[email protected]>
Reviewed-by: Miquel Raynal <[email protected]>
|
|
A carefully crafted squashfs filesystem can exhibit an extremly large
inode size and overflow the calculation in sqfs_inode_size().
As a consequence, the squashfs driver will read from wrong locations.
Fix by using __builtin_add_overflow() to detect the overflow.
Signed-off-by: Richard Weinberger <[email protected]>
Reviewed-by: Miquel Raynal <[email protected]>
|
|
A carefully crafted squashfs filesystem can exhibit an inode size of 0xffffffff,
as a consequence malloc() will do a zero allocation.
Later in the function the inode size is again used for copying data.
So an attacker can overwrite memory.
Avoid the overflow by using the __builtin_add_overflow() helper.
Signed-off-by: Richard Weinberger <[email protected]>
Reviewed-by: Miquel Raynal <[email protected]>
|
|
Since U-Boot does not support memory overcommit we can
enforce that the allocation size is within the malloc area.
This is a simple and efficient hardening measure to mitigate
further integer overflows in dlmalloc.
Signed-off-by: Richard Weinberger <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
Make sure that the new break is within mem_malloc_start
and mem_malloc_end before making progress.
ulong new = old + increment; can overflow for extremely large
increment values and memset() can get wrongly called.
Signed-off-by: Richard Weinberger <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
req is of type size_t, casting it to long opens the door
for an integer overflow.
Values between LONG_MAX - (SIZE_SZ + MALLOC_ALIGN_MASK) - 1 and LONG_MAX
cause and overflow such that request2size() returns MINSIZE.
Fix by removing the cast.
The origin of the cast is unclear, it's in u-boot and ppcboot since ever
and predates the CVS history.
Doug Lea's original dlmalloc implementation also doesn't have it.
Signed-off-by: Richard Weinberger <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
sbrk() assumes ptrdiff_t is large enough to enlarge/shrink the heap
by LONG_MIN/LONG_MAX.
So, use the long type, also to match the rest of the Linux ecosystem.
Signed-off-by: Richard Weinberger <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
Add this to the existing entry for similar boards.
Signed-off-by: Tom Rini <[email protected]>
|
|
bootstage_get_size() returns the total size of the data structure
including associated records.
When copying from gd->bootstage, only the allocation size of gd->bootstage
must be used. Otherwise too much memory is copied.
This bug caused no harm so far because gd->new_bootstage is always
large enough and reading beyond the allocation length of gd->bootstage
caused no problem due to the U-Boot memory layout.
Fix by using the correct size and perform the initial copy directly
in bootstage_relocate() to have the whole relocation process in the
same function.
Signed-off-by: Richard Weinberger <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
Evaluate the filesystem incompat and ro_compat bit fields to judge
whether the filesystem can be read or written.
For the read side only a scary warning is shown so far.
I'd love to abort mounting too, but I fear this will break some setups
where the driver works by chance.
Signed-off-by: Richard Weinberger <[email protected]>
|
|
Do not load and flush segments whose program header type is not
PT_LOAD to avoid crash happen.
bootelf will load all program headers and then flush cache.
Since PT_GNU_STACK start address is 0x0, crash happens when
it flushes PT_GNU_STACK area for target that memory cannot be
accessed from address 0x0.
crash call stack:
====================================================================
bootelf -p
|-> do_bootelf (cmd/elf.c)
|-> load_elf_image_phdr (lib/elf.c) /* Load each program header */
|-> for (i = 0; i < ehdr->e_phnum; ++i) { /* e_phnum is 2 */
|-> flush_cache(start=0, size=0) (arch/arm/lib/cache.c) （*1）
|-> flush_dcache_range(start=0, 0); (arch/arm/cpu/armv8/cache_v8.c)
|-> __asm_flush_dcache_range(start=0, 0) (arch/arm/cpu/armv8/cache.S)
|-> /* clean & invalidate data or unified cache */
|-> dc civac, x0 <<== crash happen x0:start=0
====================================================================
*1 Here the p_paddr and p_filesz of the GNU_STACK program are 0.
hello_world elf image program headers
====================================================================
Program Headers:
Type Offset VirtAddr PhysAddr
FileSiz MemSiz Flags Align
LOAD 0x0000000000000000 0x0000000080ff0000 0x0000000080ff0000
0x0000000000010390 0x0000000000010390 R E 0x10000
GNU_STACK 0x0000000000000000 0x0000000000000000 0x0000000000000000
0x0000000000000000 0x0000000000000000 RW 0x10
Section to Segment mapping:
Segment Sections...
00 .text .rodata .eh_frame
01
====================================================================
bootelf -p crash error log
====================================================================
bootelf -p 0x80f00000
"Synchronous Abort" handler, esr 0x96000147, far 0x0
elr: 000000008f6035bc lr : 000000008f66ff54 (reloc)
elr: 00000000bd9475bc lr : 00000000bd9b3f54
x0 : 0000000000000000 x1 : 0000000000000000
x2 : 0000000000000040 x3 : 000000000000003f
x4 : 0000000081000390 x5 : 00000000bd12da98
x6 : 0000000000000010 x7 : 0000000000000010
x8 : 000000000000000a x9 : 00000000bd156b30
x10: 00000000bd1526e0 x11: 0000000000000000
x12: 0000000000000004 x13: 00000000bda00000
x14: 00000000ffffffff x15: 00000000bd12dab7
x16: 00000000bd957b14 x17: 0000000000000000
x18: 00000000bd13fd90 x19: 0000000080f000b0
x20: 0000000080f00000 x21: 0000000000000002
x22: 0000000000000000 x23: 0000000000000003
x24: 00000000bd9eacc0 x25: 0000000000000000
x26: 0000000000000000 x27: 0000000000000000
x28: 00000000bd154c80 x29: 00000000bd12da70
Code: d2800082 9ac32042 d1000443 8a230000 (d50b7e20)
====================================================================
Signed-off-by: Jianfeng Zhu <[email protected]>
Reviewed-by: Jacky Cao <[email protected]>
Reviewed-by: Toyama, Yoshihiro <[email protected]>
|
|
Enable OF_UPSTREAM for Keystone boards. Remove DT files that
are now available in dts/upstream.
Signed-off-by: Andrew Davis <[email protected]>
Acked-by: Neha Malcom Francis <[email protected]>
|
|
ENOSYS should only be used when a subsystem is completely absent.
Convert its use in pinctrl-generic to ENOENT, which better reflects the
error condition (that a function/group/pin is missing).
Signed-off-by: Sean Anderson <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
Add a new defconfig file for the AM335x High Security EVM. This config
is specific for the case of SPI booting.
Signed-off-by: Andrew Davis <[email protected]>
|
|
For decoding RISC-V dumps we need to update the script.
Signed-off-by: Heinrich Schuchardt <[email protected]>
|
|
current env_set_default_vars() doesn't delete
var that are not in the imported env. hashtable
removes vars that are not in the imported
env but present in the current env only if H_NOCLEAR
flag is not set.
This change is to avoid passing H_NOCLEAR flag if
specific vars are passed to env_set_default_vars()
Without this change:
Marvell>> env default boot_mode
Marvell>>
With the change:
Marvell>> env default boot_mode
WARNING: 'boot_mode' not in imported env, deleting it!
Signed-off-by: Ravi Minnikanti <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
After applying the following patch, wget test on sandbox failed[1].
Commit: cab7867cff ("net: wget: Support retransmission a dropped packet")
Here are two reasons why the test is failed and how to fix it:
1. tcp_ack is calculated by the wrong value.
tcp_ack needs to be calculated by the received TCP payload size.
2. wget command may have a problem that HTTP response from server
must be divided into more than two packets.
In this commit, HTTP response is divided into two packets.
In addition, I fixed the HTTP response returned at the correct timing.
[1] https://lore.kernel.org/u-boot/CAFLszThEbk2Jr8OZ6Hj21wPSnJjgJhaDe037RqwHvwt1KjB3_A@mail.gmail.com/
Signed-off-by: Yasuharu Shibata <[email protected]>
Reported-by: Simon Glass <[email protected]>
|
|
Enable this so that the tests run.
Fix a few warnings in the code so that CI passes.
Signed-off-by: Simon Glass <[email protected]>
Signed-off-by: Yasuharu Shibata <[email protected]>
|
|
CONFIG_LED_BLINK and CONFIG_LED_SW_BLINK can be defined independently.
Led blinking works if any (or both) of them is enabled. Unfortunately
the led command help does not display blinking option if only
CONFIG_LED_SW_BLINK is enabled. This is definitely wrong.
This patch fix an issue.
Signed-off-by: Mikhail Kshevetskiy <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Acked-by: Alexander Dahl <[email protected]>
|
|
Some boards don't have chipselect lines for leds so cs-gpios is not
specified in the dts leading to probing error. Fix it by making
behavior similar to the one in Linux, parse num-chipselects and
if it is zero, ignore cs-gpios.
Signed-off-by: Michael Polyntsov <[email protected]>
Signed-off-by: Mikhail Kshevetskiy <[email protected]>
|
