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
System_Error has quit [Ping timeout: 264 seconds]
mraynal has quit [Remote host closed the connection]
mraynal has joined #armlinux
Lucanis has joined #armlinux
System_Error has joined #armlinux
apritzel_ has quit [Ping timeout: 246 seconds]
jclsn has quit [Ping timeout: 272 seconds]
jclsn has joined #armlinux
mvaittin has joined #armlinux
cbeznea has joined #armlinux
frieder has joined #armlinux
monstr has joined #armlinux
monstr has quit [Ping timeout: 252 seconds]
apritzel_ has joined #armlinux
apritzel_ has quit [Ping timeout: 244 seconds]
gclement has joined #armlinux
sszy has joined #armlinux
headless has joined #armlinux
sudeepholla has joined #armlinux
prabhakalad has joined #armlinux
System_Error has quit [Remote host closed the connection]
headless has quit [Quit: Konversation terminated!]
System_Error has joined #armlinux
fvincenzo has joined #armlinux
apritzel has joined #armlinux
nsaenz has quit [Ping timeout: 265 seconds]
markussp has quit [Quit: Ping timeout (120 seconds)]
markussp has joined #armlinux
derjanni has joined #armlinux
nsaenz has joined #armlinux
tomba has quit [Ping timeout: 265 seconds]
tomba has joined #armlinux
GrahamNorth has joined #armlinux
<GrahamNorth>
Hi
<rfs613>
GrahamNorth: morning ;-)
<GrahamNorth>
Hey Ralph
<derjanni>
Bonjour Ralph
<GrahamNorth>
Trying to find someone to help with ARM64 ASM
<GrahamNorth>
in vmlinux.lds.S
<rfs613>
well you're in the right channel ;-)
<GrahamNorth>
Good
<GrahamNorth>
I'm trying to set up he page tables but the 64 is quite different from the 23
<GrahamNorth>
I'm trying to set up he page tables but the 64 is quite different from the 32
<GrahamNorth>
I'm trying to set up the page tables but the 64 is quite different from the 32
GrahamNorth has quit [Quit: Client closed]
<rfs613>
that's unfortuantely something I haven't looked into myself, but others here may have
Graham-Linaro has joined #armlinux
Graham-Linaro has quit [Client Quit]
Graham-Linaro has joined #armlinux
Graham-Linaro has quit [Client Quit]
GrahamNorth has joined #armlinux
GrahamNorth has quit [Quit: Leaving]
GrahamNorth has joined #armlinux
TheCoffeMaker_ has quit [Ping timeout: 252 seconds]
TheCoffeMaker has joined #armlinux
<GrahamNorth>
I'm trying to adapt the vmlinux.lds.S in arm64 for XIP but the page table setup is very different from the arm32. Has anyone worked on this? I've already adapted the .text and some of .data sections but I'm not understanding the code for the TLB config
TheCoffeMaker has quit [Ping timeout: 244 seconds]
TheCoffeMaker has joined #armlinux
dhruvag2000 has joined #armlinux
psydroid2 has joined #armlinux
<geertu>
GrahamNorth: Just wondering, which SOC are you using?
<geertu>
IIRC, the Linux ARM64 code relies a lot on code patching, which cannot be done when using XIP.
mvaittin has quit [Ping timeout: 252 seconds]
<GrahamNorth>
The SoC is bespoke for a client. It uses 10Mb MRAM
<GrahamNorth>
based on corstone10000 (cortex-a53)
TheCoffeMaker has quit [Remote host closed the connection]
<GrahamNorth>
Geert I'm aware of alternatives patching. I'm cheating by using a 3.18 kernel for now
TheCoffeMaker has joined #armlinux
<GrahamNorth>
This was before alternatives patching was introduced
<maz>
doesn't seem very future-proof...
<ardb>
maz: using a53 in a new design?
<ardb>
that doesn't suggest being future proof is a design goal tbh
<GrahamNorth>
Yes I know. It's just to get that part worked out, then I can use the working boot code in the newest version and focus on alt patching
<GrahamNorth>
Oh you mean A53 is not future proof? Oh well that's the customers choice.
<GrahamNorth>
I'm lucky it's release date is before the alt patching
<robmur01>
IIRC I think the early pagetable code has had significant rework 2 or 3 times since 2014 - hacking on a kernel that old might be a bit of a dead end unless it's the kernel you actually intend to use
<GrahamNorth>
@geertu You worked on the page table configs?
<GrahamNorth>
e039ee4ee3fc (arm64: add alternative runtime patching, 2014-11-14)
<ardb>
GrahamNorth: it might make more sense to use a recent kernel and stub out the alternatives patching
<GrahamNorth>
@robmur01 The customer just wants something to work for a demo, so an old kernel would be OK
<ardb>
and force-enable the ones you really need at build time - there are a few A53 errata you might need
<GrahamNorth>
@ardb Tried that. It only works to stub out the boot patching. Stubbing system patching breaks it
<ardb>
ah right
<ardb>
yeah that makes sense - you'd need some global variable for the 'system patched' state
<GrahamNorth>
Yeah tried fake patching (return 0) and other things: doesn't work.
<ardb>
but in the general case, all the alternatives are formulated in a way that the unpatched state works in most cases
<ardb>
with the exception of the one you ran into, along with any a53 errata
<GrahamNorth>
Well I assume it must or why would the 3.18 work
<ardb>
so you are using disjoint code and data?
<ardb>
that seems more tricky tbh
<GrahamNorth>
For XIP? you mean .text and .data? Yeah that's the idea
<GrahamNorth>
Need to disable MMU as well
<ardb>
GrahamNorth: did you consider running the 32-bit kernel instead?
<ardb>
i assume 'userland' is going to be 32 bit anyway, given the ram constraints?
TheCoffeMaker has quit [Ping timeout: 252 seconds]
TheCoffeMaker has joined #armlinux
<GrahamNorth>
Well no. The client is running full 64 bit.
<GrahamNorth>
Yes we've talked to them and they say if they wanted 32-bit they would not have bought the A53
<maz>
ardb: no, my comment was more about relying on 3.18 as a target.
<ardb>
maz: yeah i got that :-)
<ardb>
but my point was that future proof went out the window when they picked a53, no?
<maz>
I fear this CPU will be with us forever... bit like a "new" ARM7.
<ardb>
yeah
<GrahamNorth>
The old keyboard argument
<GrahamNorth>
Honestly it doesn't make sense to get an A53 for such a small target.
derjanni has quit [Remote host closed the connection]
derjanni_ has joined #armlinux
derjanni_ is now known as derjanni
frieder has quit [Remote host closed the connection]
dhruvag2000 has quit [Quit: Connection closed for inactivity]
<robmur01>
I guess it's probably slightly cheaper than A34/A35 :)
nsaenz has quit [Remote host closed the connection]
nsaenz has joined #armlinux
nsaenz has quit [Read error: Connection reset by peer]
gclement has quit [Ping timeout: 252 seconds]
System_Error has quit [Ping timeout: 264 seconds]
markussp has quit [Ping timeout: 276 seconds]
markussp has joined #armlinux
System_Error has joined #armlinux
<arnd>
robmur01: my impression was that typically cheap SoCs end up with an A53 or A7 when they are stuck with a 28nm or 40nm process for some reason, at least I've never seen an A35 on anything larger than 22nm.
headless has joined #armlinux
nsaenz has joined #armlinux
<GrahamNorth>
Did I typo something? I mean A53
apritzel_ has joined #armlinux
<geertu>
GrahamNorth: No, I didn't work on page table configs
<arnd>
GrahamNorth: no, that's what I meant: for low-end chips one normally wants an A35 because of the lower area and power, but it's usually only available for 7nm to 22nm, compared to the bigger A53 that you can have on older 40nm and 28nm processes as well.