Both features describe an SoC family, which is the wrong granularity: a
mediatek/mt7623 board claims a display while a filogic board with a MIPI
DBI panel cannot, and rockchip claims audio for a NanoPi R2S as much as
for a Rock 5B. The audio-support and video-support packages carry that
information per image, and the modules depend on them.
Drop both features from every target that declares them.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
One interesting piece of information that has come up:
UML has come a long way with VFIO_PCI. But this functionality is
unused/disabled because the target doesn't set the PCI feature.
I think this feature could be added and the related CONFIG_VFIO
options could be selected by the kmod-vfio package.
That said, CONFIG_NO_IOMEM=y and friends will have to go.
So this can involve more work than one would expect.
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
Two patches are skipped:
- 000-v6.13-asm-generic-io.h-rework-split-ioread64-iowrite64-hel.patch
It's included. This was backported from upstream.
- 102-pseudo-random-mac.patch
The deprecated um_net driver has been removed.
You need to convert to the vector-based network driver.
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
The driver tries to access ioports (0x2f9!!) which UML doesn't have.
This causes lots of warnings to appear on boot:
| WARNING: CPU: 0 PID: 1 at lib/logic_iomem.c:188 serial8250_config_port+0x20a/0x1260
| Invalid writeqb of 0xff at address 2f9
| Modules linked in:
| CPU: 0 UID: 0 PID: 1 Comm: swapper Tainted: G W 6.12.25 #0
| Tainted: [W]=WARN
| Stack:
| Call Trace:
| [<60001000>] ? set_reset_devices+0x0/0x16
| [<60452cba>] ? serial8250_config_port+0x20a/0x1260
| [<6071e91f>] ? dump_stack_lvl+0x55/0x71
| [<606e127c>] ? _printk+0x0/0x4f
| [<60044bbb>] ? __warn+0x11b/0x120
| [<6003b5d0>] ? um_set_signals+0x0/0x40
| [<606de721>] ? warn_slowpath_fmt+0x81/0x8c
| [<606de6a0>] ? warn_slowpath_fmt+0x0/0x8c
| [<60452cba>] ? serial8250_config_port+0x20a/0x1260
(they are even recursive!)
Now, the situation gets more confusing. Because from what I can tell,
this was seemingly done intenionally. Upstream patches in related areas:
|commit ddd268c42871b78c75e12a5c28207fb481138f41
|Author: Niklas Schnelle <schnelle@linux.ibm.com>
|Date: Wed Apr 3 14:43:00 2024 +0200
|
| um: Select HAS_IOREMAP for UML_IOMEM_EMULATION
|
| In a future patch HAS_IOPORT=n will disable inb()/outb() and friends at
| compile time. UML supports these via its UML_IOMEM_EMULATION so let that
| select HAS_IOPORT and also reflect this in NO_IOPORT_MAP.
hint that there's ongoing work in this area. But unfortunately, this future
hasn't arrived yet. Once this future arrives, please nuke this patch.
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
refresh config + patches.
This includes the following changes:
- The selected x86_64 CPU is switched to generic instead of K8.
A backported patch is included from 6.13.
000-v6.13-asm-generic-io.h-rework-split-ioread64-iowrite64-hel.patch
|fixed the following build error:
| CC lib/iomap.o
|lib/iomap.c:156:5: error: no previous prototype for 'ioread64_lo_hi' [-Werror=missing-prototypes]
| 156 | u64 ioread64_lo_hi(const void __iomem *addr)
| | ^~~~~~~~~~~~~~
| [...]
Note: 102-pseudo-random-mac.patch will likely go away with the next stable.
UML is switching to a new networking infrastructure. The previous implementation
using tuntap, daemon, socket, ethertap, vde are being replaced by "vector"
transports tap, hybrid, raw, EoGRE, Eol2tpv3, fd, vde (vector!).
Please see, to checkout what will change:
<https://docs.kernel.org/virt/uml/user_mode_linux_howto_v2.html#setting-up-uml-networking>
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
This should have been fixed by an upstream patch in 5.12
|commit ed102bf2afed226703eaf85a704755bdbea34583
|Author: Randy Dunlap <rdunlap@infradead.org>
|Date: Thu Apr 15 10:13:52 2021 -0700
|
| um: Fix W=1 missing-include-dirs warnings
|
| Currently when using "W=1" with UML builds, there are over 700 warnings
| cc1: warning: ./arch/um/include/uapi: No such file or directory [-Wmissing-include-dirs]
|
| but arch/um/ does not have include/uapi/ at all, so add that
| subdir and put one Kbuild file into it (since git does not track
| empty subdirs).
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
Running the glibc-static check with GCC14 as the host compiler will fail:
Please install a static glibc package. (Missing libutil.a, librt.a or libpthread.a)
However, this error will get printed even with the required static
libraries installed when GCC14 is used.
Manually running the check exposes the real error:
<stdin>: In function ‘main’:
<stdin>:1:45: error: implicit declaration of function ‘timer_gettime’ [-Wimplicit-function-declaration]
GCC14 now errors on implicit declarations by default, so lets add the
required time.h header to fix compilation and thus the check.
Link: https://github.com/openwrt/openwrt/pull/15778
Signed-off-by: Robert Marko <robimarko@gmail.com>
UML didn't have a subtarget defined.
While this compiled fine, the patch caused funny filenames
for the generated files to appear:
- "openwrt-uml--vmlinux"
- "openwrt-uml--squashfs.img"
- "openwrt-uml--ext4.img"
- ...
since "generic" is implied anyway, let's just set it.
Fixes: d997477775 ("treewide: remove implicit SUBTARGET")
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15713
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
This option was removed from upstream kernel back in 2022.
See commits:
2d16803c562ecc644803d42ba98a8e0aef9c014e (>=6.0)
3dd33a09f5dc12ccb0902923c4c784eb0f8c7554 (>=5.15.61 backport)
Signed-off-by: Christian Svensson <blue@cmd.nu>
Add the latest default Kernel for testing.
There is no uml_watchdog userspace daemon available,
hence CONFIG_UML_WATCHDOG is disabled.
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
the UML build fails during the kernel build:
| arch/um/drivers/net_kern.c: In function 'compute_hash':
| arch/um/drivers/net_kern.c:322:1: error: the frame size of 1072 bytes is larger than 1024 bytes [-Werror=frame-larger-than=]
| 322 | }
| | ^
|cc1: all warnings being treated as errors
The compute_hash() function is added by our patch:
102-pseudo-random-mac.patch
Instead of allocating a 1024 byte buffer on the stack for the SHA1
digest input, let's allocate the data on the heap. We should be
able to do that since crypto_alloc_ahash and ahash_request_alloc
also need to allocate structures on the heap.
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
All targets are bumped to 5.15. Remove the old 5.10 patches, configs
and files using:
find target/linux -iname '*-5.10' -exec rm -r {} \;
Further, remove the 5.10 include.
Signed-off-by: Nick Hainke <vincent@systemli.org>
the 5.10 uml build currently breaks with:
/usr/bin/ld: arch/um/os-Linux/signal.o: in function `sigusr1_handler':
arch/um/os-Linux/signal.c:141: undefined reference to `uml_pm_wake'
But there's an upstream fix for this. Backport the fix
for now but also let upstream know so it finds its way
through the -stable releases.
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
This is now built-in, enable so it won't propagate on target configs.
Link: https://lkml.org/lkml/2022/1/3/168
Fixes: 79e7a2552e ("kernel: bump 5.15 to 5.15.44")
Fixes: 0ca9367069 ("kernel: bump 5.10 to 5.10.119")
Signed-off-by: Tomasz Maciej Nowak <tmn505@gmail.com>
(Link to Kernel's commit taht made it built-in,
CRYPTO_LIB_BLAKE2S[_ARM|_X86] as it's selectable, 5.10 backport)
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
Each of
- CRYPTO_AEAD2
- CRYPTO_AEAD
- CRYPTO_GF128MUL
- CRYPTO_GHASH
- CRYPTO_HASH2
- CRYPTO_HASH
- CRYPTO_MANAGER2
- CRYPTO_MANAGER
- CRYPTO_NULL2
either directly required for mac80211 crypto support, or directly
selected by such options. Support for the mac80211 crypto was enabled in
the generic config since c7182123b9 ("kernel: make cryptoapi support
needed by mac80211 built-in"). So move the above options from the target
configs to the generic config to make it clear why do we need them.
CC: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
Both CLANG_VERSION and LLD_VERISON are autogenerated runtime
configuration options, so add them to the kernel configuration filter
and remove from generic and per-target configs to keep configs clean.
Signed-off-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
Add the latest default Kernel for testing. This step is required to keep
UML in tree for the next release.
Signed-off-by: Paul Spooren <mail@aparcar.org>
The UML target been outdated for a long time. Instead of just carrying
unmaintained code we should build it again and allow people and CIs to
use it for testing.
This commit removes the `source-only` feature which disables building.
Signed-off-by: Paul Spooren <mail@aparcar.org>
This step drops support host build systems other than x86/64 to allow
two Kernel configuration in parallel. With this commit the setup follow
the config style of all other targets.
Signed-off-by: Paul Spooren <mail@aparcar.org>
The current `uml` README is terribly outdated and non of the examples
work by default. Fix that and while at it convert it to Markdown.
Signed-off-by: Paul Spooren <mail@aparcar.org>
Daniel Golle
Christian Lamparter
Shiji Yang
Rosen Penev
Stijn Tintel
John Audia
Álvaro Fernández Rojas
Robert Marko
Christian Svensson
Nick Hainke
Tomasz Maciej Nowak
Paul Spooren
Sergey Ryazanov