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
TheCoffeMaker has quit [Quit: So long and thanks for all the fish]
TheCoffeMaker has joined #armlinux
djrscally has quit [Quit: Konversation terminated!]
TheCoffeMaker has quit [Quit: So long and thanks for all the fish]
TheCoffeMaker has joined #armlinux
TheCoffeMaker has quit [Quit: So long and thanks for all the fish]
TheCoffeMaker has joined #armlinux
prabhakarlad has quit [Quit: Client closed]
dliviu has quit [Ping timeout: 264 seconds]
TheCoffeMaker has quit [Ping timeout: 250 seconds]
TheCoffeMaker has joined #armlinux
ravan__ has joined #armlinux
dliviu has joined #armlinux
ravan_ has quit [Ping timeout: 265 seconds]
dliviu has quit [Quit: Going away]
dliviu has joined #armlinux
torez has quit [Quit: torez]
prabhakarlad has joined #armlinux
apritzel_ has quit [Ping timeout: 250 seconds]
milkylainen_ has quit [Quit: Ping timeout (120 seconds)]
dliviu has quit [Ping timeout: 268 seconds]
dliviu has joined #armlinux
Pali has quit [Ping timeout: 265 seconds]
dliviu has quit [Ping timeout: 250 seconds]
dliviu has joined #armlinux
prabhakarlad has quit [Quit: Client closed]
ravan_ has joined #armlinux
ravan__ has quit [Ping timeout: 268 seconds]
ravan_ is now known as ravan
ravan has quit [Quit: Leaving]
tudorel has joined #armlinux
tudorel has quit [Ping timeout: 260 seconds]
netonaut_ has quit [Read error: Connection reset by peer]
netonaut_ has joined #armlinux
apritzel_ has joined #armlinux
milkylainen49 has joined #armlinux
milkylainen49 is now known as milkylainen_
apritzel_ has quit [Ping timeout: 268 seconds]
milkylainen_ has quit [Quit: Connection closed]
tre has joined #armlinux
nsaenz has joined #armlinux
sszy has joined #armlinux
guillaume_g has joined #armlinux
milkylainen38 has joined #armlinux
milkylainen38 is now known as milkylainen_
nsaenz has quit [Remote host closed the connection]
nsaenz has joined #armlinux
djrscally has joined #armlinux
gcl_ has joined #armlinux
prabhakarlad has joined #armlinux
<arnd>
ardb, tmlind: I see that cpu_suspend() in arch/arm/kernel/suspend.c calls the OMAP cpuidle function in SRAM with a remporary page table using identity mapping
gcl has quit [Ping timeout: 268 seconds]
<arnd>
omap34xx_cpu_suspend() then goes on to save registers to the stack, which get restored at the end of omap3_do_wfi(), but that doesn't work if the page tables does not contain the stack
matthias_bgg has joined #armlinux
tudorel has joined #armlinux
iivanov has joined #armlinux
apritzel has joined #armlinux
alexels has joined #armlinux
<tmlind>
arnd: ok, sounds like ardb already has a fix coming
<arnd>
ardb: how does this work with other CPUs accessing the overflow stack at the same time?
tudorl has joined #armlinux
<ardb>
ardb: the overflow stack is per-cpu
<ardb>
arnd:
<ardb>
sigh
<arnd>
ok
tudorel has quit [Ping timeout: 265 seconds]
tudorl is now known as tudorel
<arnd>
I misunderstood what the value of the CPU offset is here
<ardb>
arnd: that patch is broken btw -i clobber the MPIDR in r2 before it is being read
<ardb>
arnd: i'll call it per-CPU offset instead
jamestperk has quit [Read error: Connection reset by peer]
jamestperk has joined #armlinux
khilman has quit [Ping timeout: 264 seconds]
Turingtoast has joined #armlinux
khilman has joined #armlinux
<mort>
So, I currently have a system with an ADC exposed as an adc-keys node. Two questions: 1) Is it possible to get the current "key" somehow? Not as an event when it changes, but to read the current value; 2) What would be a more appropriate way to expose a potmeter that's behind an ADC?
<geertu>
mort: adc-joystick with a single axis?
gcl_ has quit [Ping timeout: 265 seconds]
<geertu>
and a suitable ABS_* linux,code?
<mort>
hmm, it seems like the adc is already exposed as an adc device
<mort>
so I can probably just remove the adc-keys thing from the dts and just use iio:device0
iivanov__ has joined #armlinux
iivanov has quit [Ping timeout: 250 seconds]
Nact has joined #armlinux
<geertu>
mort: probably you can use the iio device directly
<geertu>
adc-joystick is useful if you want it to generate input events
tudorel has quit [Quit: tudorel]
<tmlind>
ardb: so probably no point attempting to support armv6 for vmap stacks? just wondering if this is a good reason to disable armv6 by default for omap2plus_defconfig..
arnd has quit [Ping timeout: 256 seconds]
robclark has quit [Ping timeout: 256 seconds]
dianders has quit [Ping timeout: 268 seconds]
maennich has quit [Ping timeout: 256 seconds]
arnd has joined #armlinux
robclark has joined #armlinux
dianders has joined #armlinux
maennich has joined #armlinux
iivanov__ has quit [Ping timeout: 268 seconds]
ardb has quit [Ping timeout: 256 seconds]
ardb has joined #armlinux
<ardb>
tmlind: vmap stacks and irq stacks rely on THREAD_INFO_IN_TASK, which needs the TPIDR registers
<ardb>
tmlind: without that, switching between stacks is very tricky because we would need to keep multiple copies of thread_info in sync
<ardb>
which is just not worth the effort tbh
<tmlind>
ardb: ok thanks yeah i agree, will quote that for a patch for disabling armv6 by default
<ardb>
tmlind: so v6k should be ok but not plain v6
<tmlind>
yup
<tmlind>
no v6k with the old omaps
<maz>
yeah, I guess 1136 is out.
<ardb>
may it rest in peace
<maz>
or burn in hell, depending who you ask.
iivanov has joined #armlinux
<ardb>
tmlind: but weren't there v6k omap2's as well?
Pali has joined #armlinux
<tmlind>
ardb: no, just v6 for omap2420 and 2430
<ardb>
tmlind: right, so the updated 1136 with v6k extensions was never used?
<tmlind>
right
<ardb>
tmlind: time to rename it to omap3plus_defconfig then :-)
<tmlind>
heheh
<arnd>
tmlind: if you and the imx folks (ping shawnguo), I would like to actually make SMP depend on !CPU_V6 eventually, and get rid of all the hacks we have added to support that configuration
* maz
has fond memories of an H4 and an early revision of the 2420...
* tmlind
recycled all the old large proto boards years ago
<arnd>
I'm fairly sure all the other v6 we care about are v6k. There is still suport for 1136r0 on realview-eb, but I don't think we care too much about that one, as there is also one with 1176
<tmlind>
heh i think that one had metal wings i had to saw off to fit it into a rack case in the first place..
<maz>
arnd: I certainly know where to find an Integrator-CP with an 1136r0. but I guess it's better to let that one go to sleep...
<arnd>
right, I just noticed integrator as well. This is the same CPU module as on realview-eb, right?
<maz>
no, different module. the Integrator predates the Realview by a couple of years, and they switched to much higher density connectors at that time.
<arnd>
for completeness, I should probably test the qemu model for realview and integrator to see if that supports v6k by default, i.e. if this emulates an r0 or r1 version of the 1136
<maz>
and as I remember the Realview version is an FPGA, not an ASIC.
<tmlind>
arnd: not sure how much leaving out the armv6 for smp would actually save, might be worth probing the ice on the mailing list first with those patches
<arnd>
maz: Ah, I see: the EB is the fpga, and the PB is an ASIC but only for 1176 and higher, right?
<arnd>
tmlind: it's probably not a ton of code, but a lot of it is in particularly nasty areas of the kernel
<maz>
arnd: something like that (memory is fading fast). the PB was definitely a proper implementation (around 400MHz IIRC).
<ardb>
arnd: does that include smp-on-up patching as well?
<tmlind>
arnd: i recall parts of it are also needed to build armv6 with armv6k to the same kernel, but it's been years
<arnd>
ardb: it could: there is no strict need for smp-on-up patching without this, but it might still be beneficial for performance
tudorel has joined #armlinux
<tmlind>
also probably am437x needs smp-on-up patching
<arnd>
tmlind: for what?
<tmlind>
arnd: there was some cpu detection patch needed at least
<arnd>
ok, I need to write that down somewhere
<tmlind>
arnd: let me find the commit, could be i misremember
<arnd>
I do remember some hacks we have for platforms that break when a single-core SoC turns on inter-cpu coherency mode, but I thought that was a different platform, and unrelated to the patching
<tmlind>
arnd: something has to be patched dynamically for am437x
<tmlind>
arnd: i think it was commit bc41b8724f24 ("ARM: 7846/1: Update SMP_ON_UP code to detect A9MPCore with 1 CPU devices")
leming has quit [Ping timeout: 260 seconds]
<arnd>
cmpxchg() is one of the tricky bits that would benefit from ruling out this config: ldrex/strex needs v6k, irqsave/restore needs !SMP, and swp/swpb is broken on ARMv8 and for 2-byte quantities
leming has joined #armlinux
<arnd>
tmlind: ok, that is probably what I remembered after all, thanks for the link
<tmlind>
arnd: but yeah if we keep armv6 buildable without allowing smp we should be fine
<arnd>
In theory I think we could even make it possible to build a combined ARMv5/ARMv6 non-SMP kernel, but that would be a ton of work for very little practical use
<tmlind>
yeah i considered that years ago and already figured out it's not worth the hassle back then :)
iivanov has quit [Remote host closed the connection]
iivanov has joined #armlinux
torez has joined #armlinux
torez has quit [Ping timeout: 264 seconds]
torez has joined #armlinux
nsaenz has quit [Remote host closed the connection]
nsaenz has joined #armlinux
tre has quit [Remote host closed the connection]
Nact has quit [Quit: Konversation terminated!]
diederik has left #armlinux [Going to see what happens IRL, see ya!]
guillaume_g has quit [Quit: Konversation terminated!]
alexels has quit [Quit: WeeChat 3.3]
tudorel has quit [Quit: tudorel]
sszy has quit [Ping timeout: 256 seconds]
pg12 has quit [Ping timeout: 250 seconds]
pg12 has joined #armlinux
apritzel has quit [Ping timeout: 250 seconds]
prabhakarlad has quit [Quit: Client closed]
pg12 has quit [Ping timeout: 265 seconds]
pg12 has joined #armlinux
matthias_bgg has quit [Quit: Leaving]
apritzel_ has joined #armlinux
prabhakarlad has joined #armlinux
TheCoffeMaker has quit [Ping timeout: 250 seconds]
Turingtoast has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
prabhakarlad has quit [Quit: Client closed]
amitk_ has joined #armlinux
amitk has quit [Ping timeout: 264 seconds]
TheCoffeMaker has joined #armlinux
yann-kaelig has joined #armlinux
TheCoffeMaker has quit [Ping timeout: 250 seconds]
<yann-kaelig>
Hello. I would like to create a PKGBUILD for my cubietruck , but there is something I do not understand when I'm looking at the PKGBUILD source done by ArchARM. For an armv7 the source define x86_64 as arch but in the binary result I see as expected armv7h. So, where is the magic, how is this possible ?