ChanServ changed the topic of #armlinux to: ARM kernel talk [Upstream kernel, find your vendor forums for questions about their kernels] | https://libera.irclog.whitequark.org/armlinux
PobodysNerfect has quit [Ping timeout: 256 seconds]
nsaenz has quit [Ping timeout: 240 seconds]
nsaenz has joined #armlinux
sszy has joined #armlinux
<wens>
geertu: nice! thanks
nsaenz has quit [Remote host closed the connection]
apritzel has joined #armlinux
nsaenz has joined #armlinux
apritzel has quit [Ping timeout: 250 seconds]
prabhakarlad has joined #armlinux
<wens>
lore seems to have stopped ingesting mail for linux-mediatek
apritzel has joined #armlinux
bps has joined #armlinux
Perflosopher has quit [Ping timeout: 250 seconds]
Perflosopher has joined #armlinux
luispm has joined #armlinux
nsaenz has quit [Remote host closed the connection]
nsaenz has joined #armlinux
<krzk>
cool to see the blog post is useful :)
nsaenz has quit [Read error: Connection reset by peer]
nsaenz has joined #armlinux
<ukleinek>
krzk: I added a bookmark, too. (I already knew that, but swapping that info in from your blog post is easier than from the kernel documentation)
bps has quit [Read error: Connection reset by peer]
<linusw__>
the product name STM32 is pretty confusing when it is now ARMv8 64-bit...
<javierm>
geertu: I've listed a few personal examples to Thomas. With my distro dev hat I agree with his series but with my hardware enthusiast hat I think that we still need fbdev user-space support
<javierm>
geertu: as you said, most projects using SDL default to fbdev. They may support KMS but that's not the default and usually not even well tested
<geertu>
javierm: thx!
* ukleinek
thinks kconfig's "imply" is very confusing.
<ukleinek>
the docs say: This is similar to "select" as it enforces a lower limit on another symbol except that the "implied" symbol's value may still be set to n from a direct dependency or with a visible prompt.
<ukleinek>
but still if FOO "imply"s BAZ I can configure FOO=y + BAZ=m
<ukleinek>
with select this isn't possible
<geertu>
ukleinek: You mean manually configuring BAZ=M? Or is there some other dependency that prevents BAZ=y?
<geertu>
AFAIUI, imply means "select iff the target's dependencies are met"
<robmur01>
linusw__: I guess you have to think of it as an M33 SoC that happens to have some A35 coprocessors :D
<ukleinek>
geertu: that's what the quoted statement suggests, but that's wrong
<ukleinek>
geertu: look at the example and the table in Documentation/kbuild/kconfig-language.rst
<ukleinek>
if BAR=y (or there is no dependency for BAZ), FOO=y + BAZ=m is allowed
<ukleinek>
"Note: If the combination of FOO=y and BAR=m causes a link error, you can guard the function call with IS_REACHABLE()"
prabhakarlad has quit [Quit: Client closed]
Perflosopher has quit [Ping timeout: 256 seconds]
Perflosopher has joined #armlinux
bps has joined #armlinux
bps has joined #armlinux
bps has quit [Changing host]
Amit_T has joined #armlinux
nsaenz has quit [Remote host closed the connection]
nsaenz has quit [Remote host closed the connection]
DynamiteDan has quit [Excess Flood]
DynamiteDan has joined #armlinux
monstr has quit [Remote host closed the connection]
nsaenz has joined #armlinux
bps has quit [Ping timeout: 256 seconds]
<arnd>
geertu: the meaning of 'imply' changed a couple of years ago, what you remember is the old behavior
<arnd>
the way I would describe it now is that it inserts a 'default THIS_SYMBOL' in THAT_SYMBOL, so it only has an effect on 'make defconfig' and 'make olddefconfig'
<arnd>
but in a more sneaky way
<arnd>
ukleinek: IS_REACHABLE() is just as evil as imply
<arnd>
you can do almost everything you need better using correct 'depends on' and 'default' statements in Kconfig
rvalue has quit [Read error: Connection reset by peer]
<ukleinek>
arnd: so when imply's semantic changed the documenation wasn't adapted completely.
rvalue has joined #armlinux
prabhakarlad has joined #armlinux
* ukleinek
thinks that "imply" (as implemented) is little useful and he thinks that the previous semantic is more helpful. I haven't researched what was the reason to switch, so I might well miss something.
<javierm>
geertu: with default y if !DRM for FB_DEVICE you don't have any objection against Thomas' patches, right ?
<javierm>
geertu: I think that we could land that and on top I can revive the FB_CORE patches. That should give us the possibility to all the configuration matrix
heat has joined #armlinux
heat has quit [Remote host closed the connection]
heat has joined #armlinux
<arnd>
ukleinek: as far as I can tell, the documentation correctly describes what 'imply' does, but it does a bad job of steering people away from it and instead makes it sound like a feature
<ukleinek>
arnd: IMHO "This is similar to "select" as it enforces a lower limit on another symbol except that the "implied" symbol's value may still be set to n from a direct dependency or with a visible prompt." is wrong
<arnd>
I think that' sexactly what I described as a "reverse default"
<arnd>
it only changes the default of another symbol but doesn't force anything
<arnd>
def2fbffe62c0 ("kconfig: allow symbols implied by y to become m") changed the behavior, and unfortunately also pointed to IS_REACHABLE()
<arnd>
I regret I introduced IS_REACHABLE() in the first place, the intention at the time was to make all the open-coded versions of that be more readable and at least correct, but I had not expected others to use that as a feature for other drivers
* ukleinek
likes IS_REACHABLE and doesn't see the problem.
bps has joined #armlinux
bps has quit [Changing host]
bps has joined #armlinux
<arnd>
ukleinek: the problem is that it hides dependencies instead of enforcing them, so when you actually want to use a feature that is defined in a loadable module but used in built-in code, it simply doesn't work without any indication to the user that it's missing
<ukleinek>
arnd: hmm, I'd blame that on the user of IS_REACHABLE who uses it in an if condition without an else branch
<arnd>
what would you put in the #else path?
amitk has quit [Ping timeout: 268 seconds]
<arnd>
I looked for examples and found that iso9660 gets NULL from disk_to_cdi(), so if cdrom.ko is a loadable module, it ends up using the first session on a multisession cdrom, but if it's built-in, you get the last session by default
<ukleinek>
arnd: the bad thing about the iso9660 case is that the semantic is spread over two different files: Kconfig to (maybe) define dependencies and the .c file to handle the corner cases
<arnd>
ukleinek: what is the purpose of building 8250_fsl as a loadable module if nothing uses it?
<ukleinek>
arnd: that only happens on COMPILE_TEST configs
<arnd>
ukleinek: I still don't see it. I can understand wanting to have CONFIG_SERIAL_8250_FSL=m, but I don't see a point in that with CONFIG_SERIAL_OF_PLATFORM=y, if that is the only user of 8250_FSL
<ukleinek>
arnd: I also considerd depends on SERIAL_8250_FSL || !SERIAL_8250_FSL for SERIAL_OF_PLATFORM, would you like that better?
<arnd>
yes, that sounds ok, just make sure powerpc can still deal with SERIAL_8250_FSL=m, as it's referenced from built-in code there
<arnd>
this leaves it as a bool symbol but merges it into the of_platform driver
<arnd>
so they are either both built-in, or a combined loadable module
<arnd>
I guess that's broken as well if ACPI is enabled, since that would introduce a second module_init() call into the module
<ukleinek>
sending broken patches isn't that bad, I already managed to do that three times for this series *blush*
cbeznea has quit [Quit: Leaving.]
deathcamel57 has joined #armlinux
<ukleinek>
krzk: I wonder if dropping unneeded quotes in yaml files is a good idea. I thought the pro yaml editor quotes everything?!
<deathcamel57>
I have an old BSP (Linux 3.x), and I'm working on forward porting it to 6.x. Currently it's getting stuck waiting on jiffies, and I'm unsure as to whether the GIC works or not. Does anyone know how I can check if the GIC is working prior to the clock being set up?
prabhakarlad has quit [Quit: Client closed]
<krzk>
ukleinek: this process already started so just few leftovers are there
<krzk>
yamllint will complain about quotes
deathcamel57 has quit [Remote host closed the connection]
deathcamel57 has joined #armlinux
<ukleinek>
krzk: so you want to go to quoted-strings: required: only-when-needed ?
<krzk>
yep
<ukleinek>
krzk: Are you aware of the Norway-problem? That one convinced me that quoting everything is the way to go.