From: AntonMoryakov
Static analyzer detected a potential memory leak in _ehci_submit_int_msg()
when backbuffer returned from _ehci_poll_int_queue does not match
the original buffer. In this case, the allocated interrupt queue
was not destroyed before returning, leading to a memory leak.
Correcti
Hi Fabio,
Thanks for the review. Yes, this is v2 — the issue Tom pointed out
(unreachable `goto` after `return`) was fixed by replacing it with `result
= -EINVAL; goto err;`.
In v3 I’ve now removed the leftover comment. No other functional changes.
Best regards,
Anton
пт, 16 мая 2025 г. в 22:59
On Fri, May 16, 2025 at 4:43 PM wrote:
> @@ -1557,6 +1557,7 @@ static int _ehci_submit_int_msg(struct usb_device *dev,
> unsigned long pipe,
> debug("got wrong buffer back (%p instead of %p)\n",
> backbuffer, buffer);
> result = -EINVAL;
> +
From: Anton Moryakov
Ensure int_queue is properly destroyed when receiving wrong buffer
by adding error handling path. Fixes memory leak that occurred
when backbuffer validation failed.
Signed-off-by: Anton Moryakov
---
drivers/usb/host/ehci-hcd.c | 5 +
1 file changed, 5 insertions(+)
di
On Fri, May 16, 2025 at 06:42:11PM +0200, Stefan Roese wrote:
> Hi Tom,
>
> please pull this next batch of mostly cyclic patches from Rasmus:
>
Oh, and can you please update the .git/config here like you did for the
marvell PR? b4 doesn't like the git@ syntax so I had to do this manually
:)
Ap
On Fri, May 16, 2025 at 09:57:11PM +0300, Anton Moryakov wrote:
> Thanks Tom — you're right, that `goto` after `return` is unreachable. I'll
> fix that and resend with the corrected flow.
OK. Please make sure that for all of the patches you're posting, you're
not adding warnings. You can enable C
Thanks Tom — you're right, that `goto` after `return` is unreachable. I'll
fix that and resend with the corrected flow.
пт, 16 мая 2025 г. в 21:31, Tom Rini :
> On Fri, May 16, 2025 at 08:54:34PM +0300, ant.v.morya...@gmail.com wrote:
>
> > From: Anton Moryakov
> >
> > Ensure int_queue is proper
From: Anton Moryakov
Handle NULL pointer case in string formatting (%s) by printing '(null)'
instead of dereferencing NULL pointer. Makes behavior consistent with
standard printf implementations and prevents potential crashes.
Signed-off-by: Anton Moryakov "
---
lib/tiny-printf.c | 2 ++
1 file
On Fri, May 16, 2025 at 08:54:34PM +0300, ant.v.morya...@gmail.com wrote:
> From: Anton Moryakov
>
> Ensure int_queue is properly destroyed when receiving wrong buffer
> by adding error handling path. Fixes memory leak that occurred
> when backbuffer validation failed.
>
> Signed-off-by: Anton
From: Anton Moryakov
Ensure int_queue is properly destroyed when receiving wrong buffer
by adding error handling path. Fixes memory leak that occurred
when backbuffer validation failed.
Signed-off-by: Anton Moryakov
---
drivers/usb/host/ehci-hcd.c | 5 +
1 file changed, 5 insertions(+)
di
From: Anton Moryakov
Free allocated name buffer when blk_create_devicef() fails to prevent
memory leak. After successful device creation, the name ownership is
transferred to the device structure and should not be freed manually.
Signed-off-by: Anton Moryakov "
---
drivers/scsi/scsi.c | 4 +++-
-0600)
are available in the Git repository at:
g...@source.denx.de:u-boot/custodians/u-boot-watchdog
tags/u-boot-watchdog-20250516
for you to fetch changes up to 6f0a3cd7bcdd4ce25b5ad253a6ebad88d4b94fbe:
cyclic: document new guarantees for cyclic_(un)register (2025-05-16
On 07.05.25 12:58, Rasmus Villemoes wrote:
Three small patches I've been meaning to send for a while; I really
believe (as I now learnt Thomas Gleixner also does) this is the way
timer frameworks should behave.
Rasmus Villemoes (3):
cyclic: make cyclic_unregister() idempotent
cyclic: make
On Fri, May 02, 2025 at 01:38:26PM -0500, Eric Schikschneit wrote:
> This will allow arch(s) that use device tree blobs to pad the end of the
> device tree so they can be modified by board files at run time. This will
> help prevent errors such as FDT_ERR_NOSPACE from occuring.
>
> Signed-off-by:
I can then leave only
+ if (!dev) {
+ ret = -ENODEV;
+ goto err;
+ }
+
but make it higher
How do you like this option?
пт, 16 мая 2025 г. в 19:20, Simon Glass :
> Hi,
>
> On Fri, 16 May 2025 at 14:51, wrote:
> >
> > From: Anton Moryakov
> >
> > The stati
Hi,
On Fri, 16 May 2025 at 14:51, wrote:
>
> From: Anton Moryakov
>
> The static analyzer (Svace) reported
> After having been compared to a NULL value at spi-uclass.c:465,
> pointer 'dev' is passed as 1st parameter in call to function 'dev_get_flags'
> at spi-uclass.c:469, where it is dereferen
On Fri, 16 May 2025 09:36:14 +0200, Simon Glass wrote:
> https://source.denx.de/u-boot/custodians/u-boot-dm/-/pipelines/26199
> https://dev.azure.com/simon0972/u-boot/_build/results?buildId=78&view=results
>
>
> The following changes since commit 4c26de2eea6bcf5f27b13da0372d27d767cd38e3:
>
>
On Fri, 16 May 2025 13:42:02 +0200, Stefan Roese wrote:
> please pull this next batch of mostly Marvell related patches:
>
>
> - mvebu_espressobin_ultra-88f3720_defconfig: enable hwrng
> - kirkwood: Convert to DM_SERIAL for Kirkwood
From: Anton Moryakov
Added free(t.val) for T_STRING case
- Added free(t.val) in T_EOL and T_EOF cases
- Added free(t.val) in default case
- Added free(t.val) on error conditions
- Prevents memory leaks from get_token() allocations
Signed-off-by: Anton Moryakov
---
boot/pxe_utils.c | 21 +
From: Anton Moryakov
- Added buffer_delete(&manifest) before returning COMMAND_ERR.
- Prevents memory leak when file loading fails.
Signed-off-by: Anton Moryakov
---
tools/ifwitool.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/tools/ifwitool.c b/tools/ifwitool.c
ind
On Fri, May 16, 2025 at 03:50:53PM +0300, ant.v.morya...@gmail.com wrote:
> From: Anton Moryakov
>
> The static analyzer (Svace) reported
> After having been compared to a NULL value at spi-uclass.c:465,
> pointer 'dev' is passed as 1st parameter in call to function 'dev_get_flags'
> at spi-ucl
On Thu, 15 May 2025 at 17:16, Bryan Brattlof wrote:
>
> Some SoCs require a Trusted Firmware-A (TF-A) AP Trusted ROM (BL1) to
> initialize the SoC before U-Boot can run properly. Add an atf-bl1 etype
> so we can properly package BL1 into a final binary
>
> Signed-off-by: Bryan Brattlof
> ---
> M
On Sun, 4 May 2025 at 07:27, Heinrich Schuchardt
wrote:
>
>
>
> Simon Glass schrieb am Sa., 3. Mai 2025, 23:27:
>>
>> Hi Heinrich,
>>
>> On Sat, 3 May 2025 at 07:32, Heinrich Schuchardt
>> wrote:
>> >
>> > In some cases we have alternative configuration options that supply the
>> > same function
Hi Rasmus,
On Fri, 16 May 2025 at 14:54, Rasmus Villemoes wrote:
>
> Background:
>
> I have several customers that will be using a certain remote signing
> service for signing their images, in order that the private keys are
> never exposed outside that company's secure servers. This is done via
Hi Tom,
On Thu, 15 May 2025 at 22:45, Tom Rini wrote:
>
> On Thu, May 15, 2025 at 09:26:54PM +0200, Simon Glass wrote:
> > Hi Tom,
> >
> > On Wed, 14 May 2025 at 23:58, Tom Rini wrote:
> > >
> > > On Sat, May 10, 2025 at 03:42:15PM +0200, Simon Glass wrote:
> > >
> > > > Check that the standard
Hi Lukasz,
On 5/15/25 5:55 PM, Lukasz Czechowski wrote:
The RAM_ROCKCHIP_DEBUG can be used only if DEBUG_UART is
available, otherwise it won't have any effect.
I think it used to have an effect before patch 2/5 no?
If that is the case, I would recommend to reorder the patches so that
3/5 is
Hi Lukasz,
On 5/15/25 5:55 PM, Lukasz Czechowski wrote:
In case DEBUG UART is not used, define dummy macros replacing
the actual function implementations that will not be available.
This allows to compile code and avoid linker errors.
Redefine the DEBUG_UART_FUNCS macro if DEBUG UART is not avai
From: Anton Moryakov
Prevent file descriptor leaks by properly closing 'fd' and 'new_fd'
when fstat() or write() operations fail.
- Added close(fd) before return in open_for_read() if fstat() fails.
- Added close(new_fd) before return in write_image() if write() fails.
- No close needed if open(
Hi Lukasz,
On 5/15/25 5:55 PM, Lukasz Czechowski wrote:
Initialize the debug uart only in case CONFIG_DEBUG_UART is
enabled. The _debug_uart_putc is used internally by debug uart
functions, so it must be also included inside #ifdef block,
otherwise it will cause compilation warnings.
Signed-off
Hi Lukasz,
On 4/25/25 12:56 PM, Lukasz Czechowski wrote:
The HX3 is a family of USB3.0 hub controllers that comes in
different variants: CYUSB330x/CYUSB331x/CYUSB332x/CYUSB230x.
To support this hub, controlling of reset pin and two
power supplies is required.
The reset time is set to 10ms, based
On Fri, May 16, 2025 at 05:56:11PM +0300, ant.v.morya...@gmail.com wrote:
> From: Anton Moryakov
>
> Static analyzer detected a potential NULL pointer dereference when 'sym'
> pointer is passed to for_all_prompts() and other functions after being
> checked for NULL earlier in the function.
>
>
Hi Lukasz,
On 4/25/25 12:56 PM, Lukasz Czechowski wrote:
Some of the onboard hubs require multiple power supplies, so extend
the driver to support them.
The implementation is inspired by the kernel driver, as introduced
by commit [1] in the v6.10 kernel.
[1]
https://github.com/torvalds/linux/c
From: Anton Moryakov
Static analyzer detected a potential NULL pointer dereference when 'sym'
pointer is passed to for_all_prompts() and other functions after being
checked for NULL earlier in the function.
The issue could occur when called from gstr get_relations_str() with
NULL symbol pointer.
Hi Lukasz,
On 4/25/25 12:56 PM, Lukasz Czechowski wrote:
In the usb_onboard_hub_remove, the reset gpio, if available, is
freed. The pin state however, remains unchanged, as set in the
usb_onboard_hub_reset. The hub is then left enabled.
During second onboard hub probing, the hub is initially en
Hi Lukasz,
On 4/25/25 12:56 PM, Lukasz Czechowski wrote:
Currently the check in usb_onboard_hub_bind is relying on specific
compatible string for the Michrochip USB5744. Replace this with
more generic approach that will allow to add new types of devices
to the of_match table. Because the driver
On 16/05/2025 00:31, Tom Rini wrote:
Now that env_get_ip() has been removed, the include file does
not need anything from . Furthermore, include/env.h itself
includes other headers which can lead to longer indirect inclusion
paths. To prepare to remove from fix all of the
remaining places w
Some boards set fdt_high to -1 which means that the FDT is not relocated
in boot_relocate_fdt().
A comment in that function says that we assume there is space after the
existing fdt to use for padding, with the padding size set to
CONFIG_SYS_FDT_PAD
However, there is no guarantee that this space
Hi all,
On 4/9/25 5:10 PM, Quentin Schulz wrote:
This does a bit of "cleanup" by reusing constants for some FIT
properties instead of having the same string in multiple places.
Additionally, this adds a new constant for the compatible property in
FIT configuration nodes[1] which is useful for F
Hi Moteen,
On 5/16/25 1:41 PM, Moteen Shah wrote:
As per bootph schema, bootph-* property in child node should be
implied in their parent, but this feature is not implemented in
the U-Boot proper stage (before relocation) resulting in devices
not being bound because of the missing bootph-all or
On Fri, May 16, 2025 at 12:46:02PM +0200, Primoz Fiser wrote:
> Migrate to OF_UPSTREAM for phyCORE-i.MX93 since board can use upstream
> Linux kernel device-tree for phyBOARD-Segin-i.MX93.
>
> Signed-off-by: Primoz Fiser
> ---
> arch/arm/dts/Makefile | 3 +-
> arch/arm/dts/imx
From: Anton Moryakov
The static analyzer (Svace) reported
After having been compared to a NULL value at spi-uclass.c:465,
pointer 'dev' is passed as 1st parameter in call to function 'dev_get_flags'
at spi-uclass.c:469, where it is dereferenced at device.h:240.
Correct explained:
1. Added dev
From: Sumit Garg
The default DIP switch configuration on RB1/2 is to enable flashing
support via USB type-c port either using QDL or fastboot. It's just
cumbersome to get the host mode working in U-Boot via DIP switch toggle
when you need the flashing capability using the type-c port.
So instead
Background:
I have several customers that will be using a certain remote signing
service for signing their images, in order that the private keys are
never exposed outside that company's secure servers. This is done via
a pkcs#11 interface that talks to the remote signing server, and all
of that w
On 5/16/25 00:31, Tom Rini wrote:
> Now that env_get_ip() has been removed, the include file does
> not need anything from . Furthermore, include/env.h itself
> includes other headers which can lead to longer indirect inclusion
> paths. To prepare to remove from fix all of the
> remaining places
From: Anton Moryakov
The static analyzer (Svace) reported a memory leak in
bootmeth_setup_iter_order():
Dynamic memory referenced by 'order' was allocated by calloc() at
bootmeth-uclass.c:113
but could be lost when returning error codes.
This fix:
1. Adds proper error handling with goto/cleanu
On 16.04.25 08:13, Martin Schiller wrote:
Commit f6eff35b8c19 ("cmd: tlv_eeprom: handle -ENODEV error from
read_eeprom function") removed the needed 'return 0' after a successful
read. As a result, the usage message is shown when 'tlv_eeprom read' is
successfully called.
Let's fix it by adding t
t
tags/u-boot-marvell-20250516-v2
for you to fetch changes up to 7afcde09399297765945d57be4535291b2ac9549:
cmd: tlv_eeprom: return after successful read from EEPROM (2025-05-16
08:17:43 +0200)
Benjamin Schneider (1):
As per bootph schema, bootph-* property in child node should be
implied in their parent, but this feature is not implemented in
the U-Boot proper stage (before relocation) resulting in devices
not being bound because of the missing bootph-all or bootph-some-ram
property in the parent node.
To miti
Add a testcase to ensure that scan_and_prop_bootph() actually
propagates bootph-* properties to supernodes.
Signed-off-by: Moteen Shah
Reviewed-by: Simon Glass
---
tools/binman/ftest.py | 24
tools/binman/test/347_bootph_prop.dts | 21 ++
In the U-Boot pre-relocation stage, if the parent node lacks
bootph-all/bootph-some-ram property and the driver lacks a pre-reloc
flag, all of its subsequent subnodes gets skipped over from driver
binding—even if they have a bootph* property.
This series addresses the issue by scanning through all
On 17.04.25 21:37, Benjamin Schneider wrote:
This device has a hardware random number generator. Linux can
use this feature to randomize the location of the kernel in
memory for better security. However, that functionality is only
available if the bootloader firmware provides it. Enable support
f
On 06.05.25 04:58, Tony Dinh wrote:
Many Synology boards (Kirkwood, Armada 37x, and Armada 38x SoCs) use the
PIC16F1829 connecting to UART1 to provide certain functions. Mainly
for LEDs control, but it can also perform hard reset, power off, and
sending beeps.
See dts/upstream/src/arm/marvell/ar
Migrate to OF_UPSTREAM for phyCORE-i.MX93 since board can use upstream
Linux kernel device-tree for phyBOARD-Segin-i.MX93.
Signed-off-by: Primoz Fiser
---
arch/arm/dts/Makefile | 3 +-
arch/arm/dts/imx93-phyboard-segin.dts | 117 ---
arch/arm/dts/imx93-phy
Am 15.05.25 um 14:38 schrieb Manorit Chawdhry:
The commit 79d91e77f4c2 ("clk: ti: clk-k3-pll: Add additional robustness
steps to the PLL sequence") introduced a change which requires the
main_timer0 to not rely on it's own clocks which anyways was wrong.
Fix it by removing the clock dependency f
Am 15.05.25 um 14:38 schrieb Manorit Chawdhry:
The commit 79d91e77f4c2 ("clk: ti: clk-k3-pll: Add additional robustness
steps to the PLL sequence") introduced a change which requires the
main_timer0 to not rely on it's own clocks which anyways was wrong.
Fix it by removing the clock dependency f
The pcireg base is not assigned to any address, reading the
pcireg base with PS_LINKUP_OFFSET which is incorrect and
giving random values. So update the pcireg base from
devicetree so that we can read the valid PCIE link status
and PHY ready status.
Signed-off-by: Venkatesh Yadav Abbarapu
---
dr
select CMD_WGET for ARCH_QEMU. This way HTTP can be tested
on the qemu platforms in CI.
Signed-off-by: Adriano Cordova
Reviewed-by: Tom Rini
---
(no changes since v1)
cmd/Kconfig | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/cmd/Kconfig b/cmd/Kconfig
index f21d27cb27f..f
Add an http server to CI tests so that HTTP booting and
loading can be tested.
Signed-off-by: Adriano Cordova
Reviewed-by: Tom Rini
---
changes in v2:
- Remove HTTP_SERVER argument for test.py
changes in v4:
- Remove comment in .gitlab-ci.yml that was preventing
parsing
.azure-pipeli
Enable HTTP server in CI to support HTTP tests in pytest
QEMU does not emulate an HTTP server, unlike other services like DHCP or TFTP.
To enable HTTP tests during CI runs, start a simple Python HTTP server
on port 80. This allows tests that require HTTP access to run.
The HTTP server is launche
Download via both TFTP and HTTP an EFI application implementing
an HTTP client using the U-Boot EFI_HTTP_PROTOCOL implementation
as backend, and from the application then download a payload via
HTTP.
Signed-off-by: Adriano Cordova
---
test/py/tests/test_efi_loader.py | 63 +++
Add a netdump EFI app to test the U-Boot EFI network
stack.
Signed-off-by: Adriano Cordova
---
lib/efi_loader/Makefile | 1 +
lib/efi_loader/netdump.c | 851 +++
2 files changed, 852 insertions(+)
create mode 100644 lib/efi_loader/netdump.c
diff --git a/l
Fix support multiple instances of EFI_IP4_CONFIG2_PROTOCOL in
efi_ip4_config2.c. Station addresses and policies should be per
protocol instance and not global.
Signed-off-by: Adriano Cordova
---
lib/efi_loader/net/efi_ip4_config2.c | 73 +++-
1 file changed, 49 insertions
Add eth_get_ethaddr_from_dev() to retrieve the MAC address
of a specific eth udevice. The function eth_get_ethaddr()
preserves its behavior of returning the MAC address of the
current eth udevice.
Signed-off-by: Adriano Cordova
---
include/net-common.h | 1 +
net/eth-uclass.c | 11 -
efi_net.c should only [un]install protocols, leaving the
actual implementation to efi_simple_network.c and efi_pxe.c.
Signed-off-by: Adriano Cordova
---
include/efi_api.h | 28 +-
include/efi_loader.h| 10 +-
lib/efi_loader/net/Makefile |
As all the protocols moved out of efi_net.c, and EFI
net object is just an efi_object struct with network
protocols installed on it, so efi_net_obj is not
needed anymore.
Signed-off-by: Adriano Cordova
---
include/efi_loader.h| 2 +-
lib/efi_loader/net/efi_net.c|
commit 68dcbdd594d4 ("ARM: imx: Add weak default reset_cpu()")
introduced a regression that 'reset' command unable to reset
imx6ulz based BSH module's modules in the u-boot.
BSH module's imx6, imx6ulz-bsh-smm-m2.dts
Fixes: 68dcbdd594d4 ("ARM: imx: Add weak default reset_cpu()")
Signed-off-by: Mic
Rename to keep it consistent with the actial EFI protocol name
Signed-off-by: Adriano Cordova
---
lib/efi_loader/net/Makefile | 2 +-
lib/efi_loader/net/{efi_ipconfig.c => efi_ip4_config2.c} | 0
2 files changed, 1 insertion(+), 1 deletion(-)
rename lib/efi_loader/n
Add the configuration that allow to reset the board from reset
cmd
Signed-off-by: Michael Trimarchi
---
configs/imx6ulz_smm_m2_defconfig | 3 +++
1 file changed, 3 insertions(+)
diff --git a/configs/imx6ulz_smm_m2_defconfig b/configs/imx6ulz_smm_m2_defconfig
index 15a3ec5c627..436bfb78cc2 1006
efi_net.c should only [un]install protocols, leaving the
actual implementation to efi_http.c and efi_ipconfig.c
Signed-off-by: Adriano Cordova
---
include/efi_loader.h | 6 ++
lib/efi_loader/net/efi_http.c | 20 +---
lib/efi_loader/net/efi_ipconfig.c | 19 ++
The node is specified on the parent architecture u-boot.dtsi
file
Signed-off-by: Michael Trimarchi
---
arch/arm/dts/imx6ulz-bsh-smm-m2-u-boot.dtsi | 4
1 file changed, 4 deletions(-)
diff --git a/arch/arm/dts/imx6ulz-bsh-smm-m2-u-boot.dtsi
b/arch/arm/dts/imx6ulz-bsh-smm-m2-u-boot.dtsi
in
This series depends on series 456885 "test/py: enable HTTP testing".
Create lib/efi_loader/net/, implement each EFI network protocol in a
separated file under lib/efi_loader/net/ and let efi_net.c be used only
to un/install protocols. The network protocol EFI_FOO_PROTOCOL is now
implemented in lib
Created /lib/efi_loader/net/ and moved efi_net.c, efi_http.c, and
efi_ipconfig.c into it. Previously, efi_net.c contained implementations
for all supported EFI network protocols, which made the file bloated
and hard to maintain.
Now, each protocol has its own file under net/. efi_net.c is responsi
Hi Tom,
https://source.denx.de/u-boot/custodians/u-boot-dm/-/pipelines/26199
https://dev.azure.com/simon0972/u-boot/_build/results?buildId=78&view=results
The following changes since commit 4c26de2eea6bcf5f27b13da0372d27d767cd38e3:
fs.h: Switch to using rtc_def.h (2025-05-13 11:30:09 -0600)
73 matches
Mail list logo
