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
naoki has quit [Quit: naoki]
naoki has joined #armlinux
hanetzer has quit [Quit: WeeChat 3.7.1]
mraynal has quit [Read error: Connection reset by peer]
apritzel_ has quit [Ping timeout: 260 seconds]
mraynal has joined #armlinux
amitk has joined #armlinux
nsc has quit [Ping timeout: 256 seconds]
nsc has joined #armlinux
nsc is now known as Guest2700
heat has quit [Ping timeout: 256 seconds]
jclsn has quit [Ping timeout: 256 seconds]
jclsn has joined #armlinux
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #armlinux
amitk_ has joined #armlinux
biju has joined #armlinux
amitk__ has joined #armlinux
amitk_ has quit [Ping timeout: 256 seconds]
guillaume_g has joined #armlinux
derjanni_ has quit [Quit: ZNC 1.8.2+deb2build5 - https://znc.in]
derjanni has joined #armlinux
apritzel_ has joined #armlinux
apritzel_ has quit [Ping timeout: 256 seconds]
derjanni has quit [Quit: ZNC 1.8.2+deb2build5 - https://znc.in]
derjanni has joined #armlinux
derjanni_ has joined #armlinux
elastic_1 has joined #armlinux
elastic_dog has quit [Killed (cadmium.libera.chat (Nickname regained by services))]
elastic_1 is now known as elastic_dog
heat has joined #armlinux
derjanni has quit [Ping timeout: 268 seconds]
frieder has joined #armlinux
derjanni_ has quit [Quit: ZNC 1.8.2+deb2build5 - https://znc.in]
derjanni has joined #armlinux
amitk_ has joined #armlinux
derjanni has quit [Ping timeout: 272 seconds]
amitk__ has quit [Ping timeout: 272 seconds]
derjanni has joined #armlinux
amitk__ has joined #armlinux
amitk has quit [Ping timeout: 256 seconds]
iivanov has joined #armlinux
macromorgan is now known as Guest3789
Guest3789 has quit [Killed (cadmium.libera.chat (Nickname regained by services))]
macromorgan has joined #armlinux
cleger has joined #armlinux
bochenchen has joined #armlinux
cbeznea has joined #armlinux
biju has quit [Ping timeout: 272 seconds]
derjanni has quit [Quit: ZNC 1.8.2+deb2build5 - https://znc.in]
derjanni has joined #armlinux
sszy has joined #armlinux
derjanni_ has joined #armlinux
derjanni has quit [Ping timeout: 268 seconds]
headless has joined #armlinux
derjanni_ has quit [Quit: ZNC 1.8.2+deb2build5 - https://znc.in]
derjanni has joined #armlinux
<ukleinek>
linusw_: looking at the ab8500 datasheet, I wonder what is the formula backing the pwm frequency table. Do you have an idea?
<linusw_>
ukleinek: not really, I never worked close with that hardware...
derjanni_ has joined #armlinux
derjanni has quit [Ping timeout: 252 seconds]
Lucanis0 has quit [Read error: Connection reset by peer]
Lucanis0 has joined #armlinux
Lucanis0 has quit [Changing host]
Lucanis0 has joined #armlinux
<ukleinek>
linusw_: can you tell me the frequency of ab8500_sysclk?
prabhakarlad has joined #armlinux
amitk_ has quit [Ping timeout: 256 seconds]
* ukleinek
suspects it's 38400000 Hz
derjanni_ has quit [Quit: ZNC 1.8.2+deb2build5 - https://znc.in]
derjanni has joined #armlinux
amitk has joined #armlinux
derjanni has quit [Quit: ZNC 1.8.2+deb2build5 - https://znc.in]
derjanni has joined #armlinux
<ukleinek>
linusw_: I found it, it's 38400000/(4096 * (32 - i))
apritzel has joined #armlinux
<punit>
rfried: Is CONFIG_DEFERRED_STRUCT_PAGE_INIT set in your config? It should move some memory related initialization to later in the boot
amitk has quit [Ping timeout: 260 seconds]
amitk has joined #armlinux
ajfriesen6 has joined #armlinux
ajfriesen has quit [Ping timeout: 252 seconds]
ajfriesen6 is now known as ajfriesen
amitk has quit [Ping timeout: 260 seconds]
amitk has joined #armlinux
bochenchen has quit [Remote host closed the connection]
bochenchen has joined #armlinux
<rfried>
punit: thanks. checking
iivanov_ has joined #armlinux
iivanov has quit [Ping timeout: 256 seconds]
headless has quit [Quit: Konversation terminated!]
lag has quit [Ping timeout: 246 seconds]
cbeznea1 has joined #armlinux
cbeznea has quit [Read error: Connection reset by peer]
bochenchen has quit [Remote host closed the connection]
cbeznea1 has quit [Read error: Connection reset by peer]
bochenchen has joined #armlinux
cbeznea has joined #armlinux
monstr has joined #armlinux
amitk has quit [Ping timeout: 260 seconds]
<arnd>
linusw_, broonie: I have a cleanup patch for ASoC/ux500 (https://pastebin.com/0v8Rvw5U) but now I'm confused about how this currently works:
<broonie>
arnd: I have no idea if it actually works or not, I don't have any of that hardware.
<arnd>
f6c377520c26 ("ASoC: ux500_pcm: Expect different saved DMA data when obtaining from DAI store") moved it over to expecting snd_dmaengine_dai_dma_data in snd_soc_dai_get_dma_data() but ux500_pcm_request_chan(() never had the same change
<arnd>
and nothing sets the data to a 'struct ux500_msp_dma_params' when booting from DT, so it looks like it just randomly corrupts memory, unless ux500_pcm_request_chan() is never called
<broonie>
The dmaengine data should be queriable from the dmaengine API but that might've got lost somewhere.
biju has joined #armlinux
biju has quit [Client Quit]
<milkylainen>
rfried: how much ram are you talking about here? and what is a significant amount of time spent? deferred initialization is mostly useful for very large amounts of ram.
prabhakarlad has quit [Quit: Client closed]
bochenchen has quit [Remote host closed the connection]
bochenchen has joined #armlinux
<ajb-linaro>
maz I'm trying to follow the pKVM share logic, once the guest has updated the mappings in pkvm_create_mappings_locked does the host ever get signalled to say something has been shared or does it query the hypervisor?
torez has joined #armlinux
<arnd>
broonie, linusw_: I think I found it now: dmaengine_pcm_new() never calls dmaengine_pcm_compat_request_channel() since dmaengine_pcm_request_chan_of() has already registered it at that point.
<linusw_>
arnd: Lee Jones had it working at one point, admittedly using platform data and a board file. I have been trying to get it working again.
<maz>
ajb-linaro: it shouldn't need to know. either it gets a pointer to a shared page and it will be able to access it, or it randomly accesses memory it shouldn't and it will take a SIGSEGV.
<linusw_>
arnd: the thing is that I think Ux500 ASoC was slapped together in some early stage of ASOC DT and never really finished.
<linusw_>
as you can see from my draft conversion, ASOC has since grown quite a bit of abstractions in DT that should be used when implementing stuff like this.
<arnd>
linusw_: looks fairly similar to what I have done locally, so we can probably just use my patch as a start before I get dragged deeper down this rabbit hole
<linusw_>
arnd: come down the rabbit hole, we have cookies and espresso...
derjanni_ has joined #armlinux
<maz>
promises, promises...
<bjdooks>
linusw_: and definetly no grues?
derjanni has quit [Ping timeout: 248 seconds]
<linusw_>
bjdooks: only when the lights go out
<broonie>
linusw_: Pretty sure the ASoC UX500 stuff predates DT even with transition to DT coming along later but ICBW.
prabhakarlad has joined #armlinux
<linusw_>
broonie: right, but then at some point Lee Started to convert the drivers to use DT, and that was some really Flintstones type of DT bindings at that point...
<linusw_>
I don't really blame him for running out of steam, it must have been pretty daring to be one of the first in that field.
<ajb-linaro>
maz ahh ok. So for a secure guest to have a VirtIO backend we will need some sort of out of band signalling - or just probe the entire guest address space.
<ajb-linaro>
xen for example uses its xenstore to communicate shared pages to other domains
<maz>
ajb-linaro: why would you need any sort of out of band signalling? you get the MMIO exit telling you the doorbell has been written to, and you know where to look.
<maz>
virtio already work as with.
<maz>
grrr
<maz>
virtio already works as is.
<maz>
or are you trying to place the virtio *device* in a separate guest?
<maz>
there is currently no provision to support guest-to-guest sharing.
XV9 is now known as XV8|Desktop
robmur01 has quit [Remote host closed the connection]
robmur01 has joined #armlinux
iivanov_ has quit [Remote host closed the connection]
iivanov has joined #armlinux
XV8|Desktop has quit [Read error: Connection reset by peer]
<ajb-linaro>
maz I'm mainly thinking about the guest memory sharing here and trying to come up with something that can work across KVM, Xen and others. So for the KVM case it would be an option to create a new device node to represent guest memory. Arguably this is superfluous though as the KVM memfd sharing already works
<ajb-linaro>
but for VirtIO backends (vhost-user) we don't want to deal with n+1 memory devices all with their own quirks and idiosyncrasies.
<maz>
well, my take on this is that the least ABI we give, the less crap userspace has to deal with. userspace has no business randomly poking guest memory. either it was given a pointer to that memory, and it can access it, or it shouldn't even try.
frieder has quit [Remote host closed the connection]
cleger has quit [Quit: Leaving]
guillaume_g has quit [Quit: Konversation terminated!]
monstr has quit [Remote host closed the connection]
<rfried>
milkylainen: our system has 16GB or DDR. it's not much time in real hw, but in emulation. the reducing of memory minimized our boot time from 2 hours to 1 hour. so it's significant.
<milkylainen>
rfried: I wouldn't expected 16G to account for halving the kernel boot time under emulation. Maybe around 1TB. The deferred init just moves the init to a threaded init. So it's still about the same amount of work.
<punit>
Yeah - deferring just moves the work around not skip it. I wasn't sure whether the use case early debug or full system workload...
<rfried>
milkylainen yes but with few cores in parallel.
<rfried>
At least that's what the documentation states