Apart from the CRC code, where it's no problem to just compute the CRC
of the last read and read again, all other instances of read() in the
uimage code are written with the assumption that the read will not be
short.
Use read_full to satisfy this assumption.
Signed-off-by: Ahmad F
: https://github.com/barebox/barebox/issues/28
Reported-by: Rashidwi
Signed-off-by: Ahmad Fatoum
---
common/uimage.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/common/uimage.c b/common/uimage.c
index 140a08c1e426..c578edae2606 100644
--- a/common/uimage.c
+++ b/commo
We don't (yet? ;) have a Direct Rendering Manager subsystem as in Linux,
so let's rename the option to avoid confusion.
Signed-off-by: Ahmad Fatoum
---
drivers/video/Kconfig | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/video/Kconfig b/drivers/video/Kco
rt for discarding write blocks")
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
Thanks for the fix,
Ahmad
> ---
> drivers/mci/mci-core.c | 17 ++---
> 1 file changed, 14 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/mci/mci-core.c b/drivers/m
emoved, also support for YUV modes has been dropped.
>
> Tested-by: Ahmad Fatoum # rk3566 HDMI
> Signed-off-by: Sascha Hauer
> ---
> +struct driver vop2_driver = {
> + .probe = vop2_bind,
> + .name = "rockchip-vop2",
> + .of_compatible = vop2_dt_matc
Hello Peng,
Looking at the DDR setup code for i.MX8M and i.MX9 that you had contributed,
I am wondering about struct dram_cfg_param::ddrphy_trained_csr and the global
ddrphy_trained_csr array.
struct dram_cfg_param::ddrphy_trained_csr is presumably generated by the
i.MX DDR tool and it seems popu
On 13.09.24 09:59, Sascha Hauer wrote:
> The key name hint is used in barebox to find a key without iterating
> over all keys, but it's not required. Make the key name hint optional
> to keytoc.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
> ---
&
rify(), move key_name_hint to struct public_key.
>
> With this we do not need to bother the FIT image code when adding
> support for ECDSA keys.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
> ---
> common/image-fit.c | 20
&g
4 array. Fix this by printing out the key values as
> an uint64_t array.
>
> Signed-off-by: Sascha Hauer
Acked-by: Ahmad Fatoum
Cool, thanks. Now we just need a tes and see it work on big-endian MIPS :-)
Cheers,
Ahmad
> ---
> scripts/keytoc.c | 42 +
On 13.09.24 09:59, Sascha Hauer wrote:
> rsa_verify() is called from generic code, so provide a static inline
> no-op wrapper used when RSA support is disabled.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
--
Pen
y values into the binary, so this is left out.
>
> Signed-off-by: Sascha Hauer
Acked-by: Ahmad Fatoum
Some minor comments below.
> diff --git a/crypto/public-keys.c b/crypto/public-keys.c
> index dc51ef18f8..bab608be17 100644
> --- a/crypto/public-keys.c
> +++ b/crypto/public-k
gt; Also we rename CONFIG_CRYPTO_RSA_KEY to CONFIG_CRYPTO_PUBLIC_KEYS and
> CONFIG_CRYPTO_RSA_BUILTIN_KEYS to CONFIG_CRYPTO_BUILTIN_KEYS as these
> variables will be used for all types of keys, not only RSA keys.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
But two comments
On 13.09.24 09:59, Sascha Hauer wrote:
> Set the key name hint in CONFIG_CRYPTO_RSA_KEY. CONFIG_CRYPTO_RSA_KEY
> now has the form:
>
> :
>
> This is done in preparation to make CONFIG_CRYPTO_RSA_KEY a list of
> keys.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahm
't change the existing file.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
> ---
> crypto/Makefile | 6 +-
> 1 file changed, 1 insertion(+), 5 deletions(-)
>
> diff --git a/crypto/Makefile b/crypto/Makefile
> index d7a06a721d..5a484355f3 100
PTO_RSA_KEY) a few lines up.
> so all this is a complicated way of using $(CONFIG_CRYPTO_RSA_KEY)
> directly.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
> ---
> crypto/Makefile | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/
On 13.09.24 09:59, Sascha Hauer wrote:
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
> ---
> scripts/keytoc.c | 11 ---
> 1 file changed, 11 deletions(-)
>
> diff --git a/scripts/keytoc.c b/scripts/keytoc.c
> index b140160688..ecb33cbe47 100644
>
On 26.09.24 15:15, Sascha Hauer wrote:
> The Linux dw-hdmi driver is based on a driver which originally was
> written specifically for the i.MX and was converted later to a generic
> dw-hdmi driver. The barebox driver is still based on the original Linux
> driver. We just merged the dw-hdmi driver.
On 26.09.24 15:15, Sascha Hauer wrote:
> Print some more information about the framebuffer in the devinfo output:
> - x/y resolution
> - line_length
> - type of framebuffer (overlay, base plane)
> - if it's an overlay, print the base plane
>
> Signed-off-by: Sascha Hauer
> ---
> drivers/video/fb
been removed, also support for YUV modes has been dropped.
>
> Signed-off-by: Sascha Hauer
With below described adjustments:
Tested-by: Ahmad Fatoum # rk3566 HDMI
> +config DRIVER_VIDEO_ROCKCHIP_HDMI
> +bool "Rockchip HDMI driver"
> +sele
On 26.09.24 15:15, Sascha Hauer wrote:
> We currently only have dma_alloc_writecombine() for aarch32. Implement
> it for aarch64 as it is useful for mapping framebuffer memory.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
Thanks,
Ahmad
> ---
> arch/arm
On 26.09.24 15:15, Sascha Hauer wrote:
> There are several new media bus formats since we last updated
> media-bus-format.h from Linux. Update it from Linux-6.10 to
> get the new formats.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
> ---
> include/video/me
k3568")
>
> These are needed for proper HDMI clock settings on RK3568.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
> ---
> drivers/clk/rockchip/clk-rk3568.c | 8
> 1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/drivers
Hello Sascha,
On 26.09.24 15:15, Sascha Hauer wrote:
> + ret = st7789v_read_data(ctx, MIPI_DCS_GET_DISPLAY_ID, ids,
> ST7789V_IDS_SIZE);
> + if (ret)
> + return ret;
> +printf("%s: %02x %02x %02x\n", __func__, ids[0], ids[1], ids[2]);
Debugging left-over.
> +static int st778
On 26.09.24 10:22, Sascha Hauer wrote:
> On Wed, Sep 25, 2024 at 06:02:37PM +0200, Ahmad Fatoum wrote:
>> On 25.09.24 15:55, Sascha Hauer wrote:
>>> This updates lib1funcs.S from Linux
>>
>> Please mention what Linux version this was imported from.
>>
>&
Hello Sascha,
On 25.09.24 16:06, Sascha Hauer wrote:
> The old deprecated device tree binding for the spi-gpio driver uses
> "gpio-sck", "gpio-mosi" and "gpio-miso" to specify the GPIOs. Switch to
> the new binding which uses the standard GPIO property names.
>
> The old binding is still used in
On 25.09.24 16:06, Sascha Hauer wrote:
> The spi-gpio driver only supports 8bit word width. Add support for
> arbitrary word widths up to 32bit.
>
> Signed-off-by: Sascha Hauer
With below issue addressed:
Reviewed-by: Ahmad Fatoum
> static int gpio_spi_setup(struct
Hello Sascha,
On 25.09.24 15:55, Sascha Hauer wrote:
> Until now there has been no assembler optimized version of memmove() for
> ARM. Add this from Linux for both ARM32 and ARM64. This also updates
> memcpy() for ARM64 from Linux.
>
> Signed-off-by: Sascha Hauer
Reviewed-b
On 25.09.24 16:06, Sascha Hauer wrote:
> spidelay() currently only is a no-op dummy function. Actually delay in
> this function to avoid spi-gpio being faster than specified.
>
> Signed-off-by: Sascha Hauer
> ---
> drivers/spi/gpio_spi.c | 5 -
> 1 file changed, 4 insertions(+), 1 deletion(-
On 25.09.24 15:55, Sascha Hauer wrote:
> This updates the assembler optimized memcpy() and memset() functions
> from Linux.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
with a small nitpick:
> /* SPDX-License-Identifier: GPL-2.0-only */
> -/* SPDX-FileC
On 25.09.24 16:34, Sascha Hauer wrote:
> struct fbc_priv has rows and cols fields, but they do not actually
> store the rows and columns, but instead the actual values minus one.
> Change this to make the code better understandable.
>
> Signed-off-by: Sascha Hauer
Reviewed-b
On 25.09.24 16:34, Sascha Hauer wrote:
> When we have a background image we want the background to actually shine
> through, so set the alpha value of the background to transparent and not
> opaque.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
I wonder if this cou
> At least this was the intention. The function sets the alpha value
> correctly only when it's not fully opaque. Fix this by always setting
> the alpha channel when it exists.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
> ---
> lib/gui/graphic_uti
On 25.09.24 16:34, Sascha Hauer wrote:
> the splash command has an option for setting a background colour. This
> doesn't always fill the whole screen as expected.
>
> struct fb_info has a line_length field that specifies the length in
> bytes of a single line. That length may be bigger than the a
On 25.09.24 16:34, Sascha Hauer wrote:
> Do not render the parts of the image that are outside the screen
> boundaries to prevent memory corruptions.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
> ---
> lib/gui/graphic_utils.c | 6 ++
> 1 file
fault_memcpy already.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
> ---
> lib/string.c | 10 --
> 1 file changed, 8 insertions(+), 2 deletions(-)
>
> diff --git a/lib/string.c b/lib/string.c
> index 374f326143..98dd3cffdd 100644
> --- a/lib/st
when we build
for
CONFIG_THUMB2_BAREBOX. I suggest, we remove this line and see if we even have
any
warnings.
With this addressed:
Reviewed-by: Ahmad Fatoum
Cheers,
Ahmad
--
Pengutronix e.K. | |
Steuerwalder Str. 21 | http:/
On 25.09.24 15:55, Sascha Hauer wrote:
> This updates io-readsb.S, io-readsl.S, io-readsw-armv4.S, io-writesb.S,
> io-writesl.S and io-writesw-armv4.S from Linux.
Likewise, please add version.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
> ---
> arch/arm
On 25.09.24 15:55, Sascha Hauer wrote:
> This updates findbit.S from Linux
Please mention version.
>
> Signed-off-by: Sascha Hauer
> ---
> arch/arm/lib32/findbit.S | 243
> ++-
> 1 file changed, 94 insertions(+), 149 deletions(-)
>
> diff --git a/a
On 25.09.24 15:55, Sascha Hauer wrote:
> This updates lib1funcs.S from Linux
Please mention what Linux version this was imported from.
> - moveq pc, lr
> + reteq lr
huh, what kind of instruction is reteq?
Cheers,
Ahmad
--
Pengutronix e.K. |
ng # PXA3xx
> | Acked-by: Stefano Stabellini # Xen
> | Tested-by: Uwe Kleine-König # ARMv7M
> | Tested-by: Simon Horman # Shmobile
> | Signed-off-by: Russell King
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
> ---
> arch/arm/cpu/
d 'push' instruction mnemonic. That allows assembler.h to be
> | included in .S files that use 'push' instruction.
> |
> | Suggested-by: Will Deacon
> | Signed-off-by: Victor Kamensky
> | Acked-by: Nicolas Pitre
> | Signed-off-
On 25.09.24 15:55, Sascha Hauer wrote:
> The optimized versions are there for ARM32, but since 0d53f3c584a2
> ("arm: use asm-generic/io.h") they were no longer used. Activate
> them again.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
> ---
> arch/a
On 13.09.24 09:59, Sascha Hauer wrote:
> In case of an error in gen_key() return an error code so that the keytoc
> tool returns an error.
>
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
> ---
> scripts/keytoc.c | 2 +-
> 1 file changed, 1 insertion(+), 1
On 13.09.24 09:59, Sascha Hauer wrote:
> Generating dts snippets for ECDSA keys does not work properly,
> there are various endianess problems. As we do not need this currently
> for barebox drop the support for now.
>
> Signed-off-by: Sascha Hauer
Acked-by: Ahmad Fatoum
&
keep the order elsewhere in the file, where
x86 checks are done before those for ARM. Both can't be active at the
same time, so no functional change expected.
Signed-off-by: Ahmad Fatoum
---
efi/payload/image.c | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/e
malloc(), free() and friends are declared in for hosted
implementations. Given that we have the header in barebox too, we should
define malloc(), free() as users porting userspace code may expect it
there.
Signed-off-by: Ahmad Fatoum
---
include/stdlib.h | 1 +
1 file changed, 1 insertion
With recent restructuring of the headers, kmalloc_array was no longer
indirectly included, which broke the build.
Add the missing headers and while at it switch over to normal malloc,
because kmalloc_array() uses a bigger alignment than necessary.
Signed-off-by: Ahmad Fatoum
---
lib/jsmn.c | 4
On 06.09.24 12:40, Sascha Hauer wrote:
> This series implements ECDSA signature verification for FIT images.
> The ECDSA code itself is taken from the Kernel. Currently only supported
> way to specify a ECDSA key is to compile it into the binary using
> CONFIG_CRYPTO_ECDSA_KEY, taking it from a dev
ates them first, so this change looks sensible.
> Signed-off-by: Sascha Hauer
Reviewed-by: Ahmad Fatoum
> ---
> crypto/public-keys.c | 10 +-
> scripts/keytoc.c | 12
> 2 files changed, 9 insertions(+), 13 deletions(-)
>
> diff --git a/crypto/public
On 06.09.24 12:40, Sascha Hauer wrote:
> This encapsulates struct rsa_public_key into a struct public_key. So far
> RSA keys are the only supported key type. With adding ECDSA keys we need
> a container struct so that we can add ECDSA keys using the same
> mechanisms.
>
> Also we rename CONFIG_CRY
Hi Sascha,
On 06.09.24 12:40, Sascha Hauer wrote:
> This adds the missing pieces to make CONFIG_CRYPTO_PUBLIC_KEYS
> a list of keys rather than a single key only.
>
> We have to remove the quotes from the key argument in cmd_keys to
> pass the keys as multiple arguments to keytoc. While at it the
Hello Sascha,
On 06.09.24 12:40, Sascha Hauer wrote:
> Set the key name hint in CONFIG_CRYPTO_RSA_KEY. CONFIG_CRYPTO_RSA_KEY
> now has the form:
>
> :
>
> This is done in preparation to make CONFIG_CRYPTO_RSA_KEY a list of
> keys.
IIUC, existing users that do make oldconfig will:
- either lo
When network booted, these properties may be missing and informing the
user about it arguably just clutters the log.
Thus bump down the log level to debug.
Signed-off-by: Ahmad Fatoum
---
arch/arm/boards/raspberry-pi/rpi-common.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff
The function loads both the built-in env and the barebox.env file on the
SD-Card's boot partition. Failure to mount the SD-Card should still
result in the built-in env being loaded, therefore move it to the start
of the function.
Signed-off-by: Ahmad Fatoum
---
arch/arm/boards/raspberry-p
extra error code
results in no loss of information.
Signed-off-by: Ahmad Fatoum
---
arch/arm/boards/raspberry-pi/rpi-common.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/arch/arm/boards/raspberry-pi/rpi-common.c
b/arch/arm/boards/raspberry-pi/rpi-common.c
index 7a3911e
We only support providing PSCI on a single SoC, the i.MX7. Having the
option in other configuration is thus confusing.
Therefore hide the symbols when not building for the i.MX7. Using PSCI
as consumer is a separate symbol (ARM_PSCI_CLIENT) and should be
unaffected.
Signed-off-by: Ahmad Fatoum
Adding X - 1 and then dividing by X is effectively rounding up, so let's
use the macro to make this clearer.
No functional change.
Signed-off-by: Ahmad Fatoum
---
lib/fonts/fonts.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/lib/fonts/fonts.c b/lib/fonts/fonts.c
The file has references to DIV_ROUND_CLOSEST_ULL and struct device_node,
which so far only worked because of headers previously included.
Fix this to make the header self-contained.
Signed-off-by: Ahmad Fatoum
---
include/pwm.h | 2 ++
1 file changed, 2 insertions(+)
diff --git a/include
idated, so the CPU sees what was written by the device.
It's therefore sensible to poison access to a buffer while owned by the
device and only unpoison it after ownership has been transferred to the
CPU.
Signed-off-by: Ahmad Fatoum
---
drivers/dma/debu
A
sync").
Such repeated sync's are usually a result of mapping a buffer and then
calling dma_sync_single_for_device on it. This is unnecessary and warrants
a renewed look at the DMA management scheme, which this warning hopefully
accomplishes.
Signed-off-by: Ahmad Fatoum
---
a01ae0>] (0xa01ae0) from [<00a01360>] (0xa01360)
Memory state around the buggy address:
40419e00: 00 00 00 00 00 00 00 00 02 fc ff ff ff ff fc 00
40419e80: 00 00 00 fc 00 00 00 00 04 fc ff ff ff ff fc fc
>40419f00: f8 f8 f8 f8 f8 f8 f8 f8 f8 f8 f8 f
#x27;s
revert commit 1ad4b32702ddc3b801e8cd35441d2a9b3a41bb19 first to have
a clean slate.
Fixes: 1ad4b32702dd ("dma: debug: detect repeated DMA sync")
Signed-off-by: Ahmad Fatoum
---
drivers/dma/debug.c | 22 ++
1 file changed, 2 insertions(+), 20 deletions(-)
diff --git a
ect double poisons/unpoisons
that should not occur if the DMA API is correctly used.
Signed-off-by: Ahmad Fatoum
---
include/linux/kasan.h | 2 ++
lib/kasan/generic.c | 37 +
2 files changed, 39 insertions(+)
diff --git a/include/linux/kasan.h b/include/lin
Give all errata workaround functions a kernel-doc listing the title of
the erratum and what CPUs it affects for documentation purposes.
Signed-off-by: Ahmad Fatoum
---
arch/arm/include/asm/errata.h | 83 +++
1 file changed, 83 insertions(+)
diff --git a/arch/arm
he warning
completely felt wrong. Looking more into it now, we can mark the
.barebox_rodata section in the linker file snippet we have in sandbox as
READONLY and the warning disappears, so let's do that instead.
Signed-off-by: Ahmad Fatoum
---
v1 -> v2:
- add fallback for earlier binuti
It's possible for Linux notifiers to be removed from within the notifier
callback. Allow the same within barebox to avoid subtle bugs.
Signed-off-by: Ahmad Fatoum
---
lib/notifier.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/lib/notifier.c b/lib/notifier.c
The header has evolved upstream since we last imported
it. Sync it with Linux to make available some more functions for use in
barebox.
Signed-off-by: Ahmad Fatoum
---
include/linux/math.h | 173 ---
lib/math/Makefile| 1 +
lib/math/int_pow.c
Linux defines int_sqrt in linux/math.h and has the implementations in
lib/math. We have both these locations, so move the existing code over
to simplify future sync of math functions.
Signed-off-by: Ahmad Fatoum
---
commands/fbtest.c | 2 +-
include/int_sqrt.h| 8
one commit that we don't take over though as that doesn't apply
to barebox yet:
321648455061 kbuild: use obj-y instead extra-y for objects placed at the head
Signed-off-by: Ahmad Fatoum
---
scripts/clang-tools/gen_compile_commands.py | 50 -
1 file changed, 29 i
Hello Rouven,
On 03.09.24 11:37, Rouven Czerwinski wrote:
> In gcc 13.3 there is a warning that start may be used unitialized:
>
> include/linux/iopoll.h:42:21: warning: ‘start’ may be used uninitialized
> [-Wmaybe-uninitialized]
> 42 | is_timeout(start, ((timeout_us)
The biggest font we have, VGA8x16, looks quite small on higher density
displays like the 15 inch 1080p display used for the barebox demo at
FrOSCon. To make it easier to discern the text, import a bigger font
from Linux.
Signed-off-by: Ahmad Fatoum
---
lib/fonts/Kconfig |9 +
lib
: Ahmad Fatoum
---
drivers/video/fbconsole.c | 17 +
1 file changed, 13 insertions(+), 4 deletions(-)
diff --git a/drivers/video/fbconsole.c b/drivers/video/fbconsole.c
index 6c85e8e06a8c..cec1afec56cb 100644
--- a/drivers/video/fbconsole.c
+++ b/drivers/video/fbconsole.c
@@ -117,7
Guide users into the right direction by mentioning what config.txt
key is needed to start barebox in 64-bit mode.
Signed-off-by: Ahmad Fatoum
---
Documentation/boards/bcm2835.rst | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/Documentation/boards/bcm2835.rst b
, have the core only report depress event for
previously pressed keys.
Signed-off-by: Ahmad Fatoum
---
drivers/input/input.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/drivers/input/input.c b/drivers/input/input.c
index 1a4792935114..e5509bf90d3e 100644
--- a/drivers
We have math/ in the same directory, so having a math.c file is
a bit confusing. Rename the file to a more descriptive arith.c.
Signed-off-by: Ahmad Fatoum
---
lib/Makefile| 2 +-
lib/{math.c => arith.c} | 0
2 files changed, 1 insertion(+), 1 deletion(-)
rename lib/{mat
As part of the measures to factor out stuff from common.h into dedicated
headers, move out STACK_ALIGN_ARRAY. This macro doesn't exit in Linux,
but the file is fitting.
We don't need to explicitly include it, because it's being included by
.
Signed-off-by: Ahmad Fatoum
---
i
to point at
correct version")
Signed-off-by: Ahmad Fatoum
---
test/Containerfile | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/test/Containerfile b/test/Containerfile
index 1c4932f56a64..5c49164d33b6 100644
--- a/test/Containerfile
+++ b/test/Containerf
[1] for
details.
[1]: https://github.com/pytest-dev/pytest/issues/2704
Suggested-by: Bastian Krause
Signed-off-by: Ahmad Fatoum
---
conftest.py | 19 +++
1 file changed, 15 insertions(+), 4 deletions(-)
diff --git a/conftest.py b/conftest.py
index 56d4b512a912..5639b7e22e
rc3, but I submitted a patch to
restore serial@9000 as default. Until that's merged, let's just
activate all consoles after describing them in the device tree.
Signed-off-by: Ahmad Fatoum
---
Optimally, we would start consuming the device tree supplied by QEMU,
but this should be good en
Hello,
On 18.08.24 19:09, Ahmad Fatoum wrote:
> Hello Stafford,
>
> On 18.08.24 09:14, Stafford Horne wrote:
>> On Wed, Aug 14, 2024 at 06:14:35PM +0200, Ahmad Fatoum wrote:
>>> The newest toolchain bump breaks OpenRISC start-up under QEMU: The console
>>> is
&g
arguments.
Signed-off-by: Ahmad Fatoum
---
MAKEALL | 24
1 file changed, 12 insertions(+), 12 deletions(-)
diff --git a/MAKEALL b/MAKEALL
index df318c6d123c..68ebe6ef98a3 100755
--- a/MAKEALL
+++ b/MAKEALL
@@ -281,20 +281,20 @@ then
REGEX="*"
fi
-if [
once.
The benefit of this is that we can have a single command to get QEMU
running on a target, e.g.:
scripts/container.sh ./MAKEALL test/arm/multi_v8_defconfig.yaml
Signed-off-by: Ahmad Fatoum
---
MAKEALL | 102
1 file changed, 96 insert
t variable to a more descriptive exitcode.
Signed-off-by: Ahmad Fatoum
---
MAKEALL | 13 +
1 file changed, 9 insertions(+), 4 deletions(-)
diff --git a/MAKEALL b/MAKEALL
index 861d397814d9..df318c6d123c 100755
--- a/MAKEALL
+++ b/MAKEALL
@@ -8,7 +8,7 @@ warnings_list="
pytest, so let's rewrite the documentation to use them.
There's still some features remaining that only emulate.pl supports,
once these are present in MAKEALL/pytest, we can drop it too.
Signed-off-by: Ahmad Fatoum
---
Documentation/boards/emulated.rst | 85 +--
level higher as described in:
https://github.com/pytest-dev/pytest/issues/5822#issuecomment-697331920
Signed-off-by: Ahmad Fatoum
---
test/conftest.py => conftest.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
rename test/conftest.py => conftest.py (99%)
diff --git a/test/co
let's have /chosen/stdout-path point at it.
Signed-off-by: Ahmad Fatoum
---
arch/mips/dts/black-swift.dts | 4
arch/mips/dts/dlink-dir-320.dts | 4
arch/mips/dts/img-ci20.dts | 4
arch/mips/dts/loongson-ls1b.dts | 4
arch/mips/dts/rzx50.dts | 4
5 fil
The semihosting DEBUG_LL support was tested on ARM64. Trying it out on
ARM32 leads to a hanging system, because the DEBUG_LL handler used the
wrong instruction (breakpoint) instead of what the normal semihosting
implementation uses (supervisor call).
Signed-off-by: Ahmad Fatoum
---
arch/arm
We have a qemu-maltael_defconfig, which is generated by Kbuild by
merging qemu-maltael_defconfig with a Kconfig fragment that switches
to little endian. Let's make use of it, instead of depending on
the runner kconfig_add key, which is only evaluated by emulate.pl.
Signed-off-by: Ahmad F
y pointing stdout-path at the first console.
Signed-off-by: Ahmad Fatoum
---
arch/mips/dts/qemu-malta.dts | 2 ++
1 file changed, 2 insertions(+)
diff --git a/arch/mips/dts/qemu-malta.dts b/arch/mips/dts/qemu-malta.dts
index c2bc03dd4631..6cf7513ff5fa 100644
--- a/arch/mips/dts/qemu-malta.dts
From: Ahmad Fatoum
The file is unreferenced anywhere and it doesn't differ from the default
__barebox_arm_head anyway, except by two additional words with value
0x, which shouldn't matter.
It thus seems that the file wasn't needed when it was first added and
it's s
MAKEALL.
Fixes: 6d4c6b01f4c4 ("ci: container: downgrade OpenRISC toolchain from 14.2.0
to 13.1.0")
Signed-off-by: Ahmad Fatoum
---
This is hopefully the last patch needed to make CI green again.
---
test/Containerfile | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --
We currently, only rebuild the container when its Containerfile changes.
The container build is described by the workflow in container.yml,
so Github changes should also rebuild the container on changes to this
file.
Signed-off-by: Ahmad Fatoum
---
.github/workflows/container.yml | 1 +
1 file
ompilation for 32-bit")
Signed-off-by: Ahmad Fatoum
---
v2 -> v3:
- make commit apply on top of v1, which has already been applied.
v1 -> v2:
- fix CI x86_64 breakage, by keeping the packed/aligned specifiction
as is, but changing it to 32-bit on 32-bit systems.
---
include/ac
Hello Stafford,
On 18.08.24 09:14, Stafford Horne wrote:
> On Wed, Aug 14, 2024 at 06:14:35PM +0200, Ahmad Fatoum wrote:
>> The newest toolchain bump breaks OpenRISC start-up under QEMU: The console is
>> now completely silent. Steps to reproduce the hang:
>>
>
From: Ahmad Fatoum
Build test on x86 fails due to pointer cast and because of forcing the
bigger alignment on the flexible array.
The problematic pointer cast is for the XSDT, which is 64-bit-only, so
let's fix the cast and return an error if we somehow see an XSDT on
32-bit.
For the fle
Hello Tomas,
On 15.08.24 07:53, Tomas Marek wrote:
> Hello Ahmad,
>
> Thanks for maintenance of the Barebox EFI.
>
> On Wed, Aug 14, 2024 at 11:24:23AM +0200, Ahmad Fatoum wrote:
>> I2C message length is 16 bit, so doing a 64-bit division is overkill.
>> Let's
-by: Ahmad Fatoum
---
include/linux/math.h | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/include/linux/math.h b/include/linux/math.h
index acbffa96e6eb..f5d5cc714c4e 100644
--- a/include/linux/math.h
+++ b/include/linux/math.h
@@ -23,7 +23,8 @@
#define DIV_ROUND_DOWN_ULL(ll
ision. Make use of this to fix compilation for
32-bit.
Cc: Tomas Marek
Signed-off-by: Ahmad Fatoum
---
This replaces "i2c: efi: avoid 64-bit division"
---
drivers/i2c/busses/i2c-efi.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/i2c/busses/i2c-efi.c b/drivers
The early startup code is written in assembly, so the C DEBUG_LL
implementation isn't as useful, but it can still come in handy as
debugging aid, so let's add it.
Signed-off-by: Ahmad Fatoum
---
arch/openrisc/Kconfig| 1 +
arch/openrisc/include/asm/debug
The functions were changed to take a base parameter, so they can be used
with PBL console instead of just DEBUG_LL. Update the comment to reflect
the current state of the code.
Fixes: e64990099c8f ("debug_ll ns16550: Add base address argument to register
functions")
Signed-off-by: Ah
1 - 100 of 1413 matches
Mail list logo
