Tartarus changed the topic of #u-boot to: SOURCE MOVED TO https://source.denx.de/u-boot/u-boot.git / U-Boot v2024.10, v2025.01-rc4 are OUT / Merge Window is CLOSED, next branch is OPEN / Release v2025.01 is scheduled for 06 January 2025 / Channel archives at https://libera.irclog.whitequark.org/u-boot
naoki has joined #u-boot
ikarso has quit [Quit: Connection closed for inactivity]
alpernebbi has quit [Ping timeout: 248 seconds]
zibolo has quit [Ping timeout: 260 seconds]
zibolo has joined #u-boot
alpernebbi has joined #u-boot
rvalue- has joined #u-boot
rvalue has quit [Ping timeout: 276 seconds]
rvalue- is now known as rvalue
xroumegue has quit [Ping timeout: 248 seconds]
xroumegue has joined #u-boot
Daanct12 has joined #u-boot
mmu_man has quit [Ping timeout: 260 seconds]
prabhakalad has quit [Ping timeout: 244 seconds]
prabhakalad has joined #u-boot
vardhan has joined #u-boot
haritz has quit [Ping timeout: 248 seconds]
gsz has joined #u-boot
goliath has joined #u-boot
frieder has joined #u-boot
monstr has joined #u-boot
gsz has quit [Ping timeout: 244 seconds]
ikarso has joined #u-boot
monstr has quit [Ping timeout: 252 seconds]
vardhan_ has joined #u-boot
vardhan has quit [Ping timeout: 244 seconds]
vardhan__ has joined #u-boot
vardhan_ has quit [Ping timeout: 272 seconds]
warpme has joined #u-boot
wooosaiiii has joined #u-boot
ldevulder has joined #u-boot
rockosov has joined #u-boot
monstr has joined #u-boot
mckoan|away is now known as mckoan
gsz has joined #u-boot
sszy has joined #u-boot
Guest2959 has joined #u-boot
Guest2959 has left #u-boot [Leaving]
gsz has quit [Ping timeout: 265 seconds]
vardhan__ has quit [Remote host closed the connection]
vardhan__ has joined #u-boot
macromorgan has quit [Ping timeout: 248 seconds]
macromorgan has joined #u-boot
vardhan__ has quit [Remote host closed the connection]
warpme has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
warpme has joined #u-boot
gsz has joined #u-boot
Jones42 has joined #u-boot
slobodan_ has joined #u-boot
slobodan__ has joined #u-boot
slobodan_ has quit [Ping timeout: 244 seconds]
<pivi> uff, from what I can see all the imx8m with >=4GB RAM are currently broken in U-Boot master, not just verdin.
<apalos> pivi: any of the changes that were proposed solves it ?
<pivi> if we do not get to a solution we agree with in short term, probably is better to just revert that commit and have a proper solution for upcoming merge window
<pivi> apalos: please see my last email
<pivi> apalos: the trivial fix, will introduce regression somewhere else
<apalos> yea reading those sec
<apalos> why so the question why ram_top was defined as 0xffffffff to begin with
<pivi> see commit e27bddff4b97 ("imx8m: Restrict usable memory to space below 4G boundary").
<pivi> frieder: ^^
<apalos> ok so whso what we havnt answered is why some allocation with top_ram defined at 0xffffffff
<apalos> and it's EFI one iirc looking at your trace
<pivi> and mmc
<apalos> which doesnt make too much sense tbh, the allocations happens way below ram_top to begin with
mmu_man has joined #u-boot
<apalos> pivi: can you paste the original error again p,s
<apalos> ah nvm I got it in mail
<apalos> so pivi can you do me a favour since I dont have access to that board?
<apalos> Apply this, beacuse it's the only place that makes sense to show the error
<apalos> All the other messages are calloc failures, AFAICT, so let's try to see if thgis is EFI specific or not
BobBeck3 has quit [Quit: The Lounge - https://thelounge.chat]
BobBeck3 has joined #u-boot
<pivi> apalos: yes, if you need more test just reply to the email thread, joao is going to be able to test easier than myself
dsimic has quit [Ping timeout: 245 seconds]
dsimic has joined #u-boot
warpme has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<frieder> pivi: e27bddff4b97 ("imx8m: Restrict usable memory to space below 4G boundary") is a workaround for some IP cores (FEC, USDHC, etc.) which don't support direct access to the space above 0xffffffff.
<apalos> frieder: yes I was about to say the same thing,
<apalos> So looking at it, this is a platform definition bug
<apalos> We'll send a reply explaining what's going on, but that ram_top @ 0xffffffff is what triggers it,
<apalos> because we are trying to protect the relocated u-boot which ends at @ 0x100000000
<apalos> and since ram_top is defined lower, the rextra byte has already been reserved as 'dont overwrite'
<apalos> We should have some code printing it's a platform bug, but since we are close to the release, let's fix this and send an update post 2025.01
<frieder> apalos: This is the relevant change that introduced the current ram_top value: https://source.denx.de/u-boot/u-boot/-/commit/74f88b72219e178dc05d8c81e21048212b04cd09
<frieder> If the (intermediate) fix is to change that value, then someone should post a patch and mention this commit in the Fixes tag.
warpme has joined #u-boot
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #u-boot
<apalos> frieder: no that's kind of irrelevant
<apalos> The actual problem pre-existed AFAICT, it juist unearthed with the lmb changes
<apalos> So we now use LMB to reserve the u-boot .text .data etc regions
<apalos> and the ram_top is supposed to be the next address,
<apalos> So when we define ram_top as ffffffff the lmb code reserves some memory and then the u-boot reservation fails
<apalos> that why it works if you bump it to 100000000,
<apalos> Now I would argue this is generic platform port bug and we need a to print the error, but I'll have a look post the release
<apalos> For now we should just fix the imx platforms that are affected
<apalos> Tartarus: since we close to the release, let's have a chat on how to sort this
<apalos> frieder: ah sorry the patch you pasted *is* relevant
<apalos> remember why you set it 0xffffffff
<apalos> the whole lmb code assumes the last usable address is included, so this needs to be 0x100000000
<apalos> frieder: sorry I thought you pasted the older commit from that mail, limiting it to 0x80000000
<apalos> so yea, you are right that needs a fixes thag
<apalos> tag*
<pivi> apalos: Tartarus: I do not see us being able to help mure more in the next weeks, everybody in my team is going to be off or super busy as the last standing man
<pivi> but the issue should be super clear now
<pivi> and it's generic on imx8m, nothing specific to our board
<pivi> s/mure/much/
<apalos> pivi: the fix should be an oneliner + the explanation in mail
<apalos> If noone from your team can do that, I can probably send it
Daanct12 has quit [Ping timeout: 265 seconds]
steffen[m] has joined #u-boot
mripard has joined #u-boot
mmu_man has quit [Ping timeout: 265 seconds]
<steffen[m]> Hi, we've been looking at improving boot time on a xilinx zynqmp device with upstream u-boot (~8 seconds for u-boot). Much of the time is spent searching for things in device tree before caches are enabled. Right now we're looking at reordering the dts so we find things quicker, or moving things behind the cache setup. (serial port init without caches takes ~1.3 seconds, spl takes less than that for everything). Is...
<steffen[m]> ... anyone having similar issues? Maybe we cold move dram/cache init to be earlier in general?
monstr has quit [Ping timeout: 252 seconds]
<marex> steffen[m]: something like https://source.denx.de/u-boot/u-boot/-/commit/ac9153c74f3 might help with DT traversal performance ?
<marex> apalos: how do you fit that bit 32 into 32bit variable ? :)
<marex> apalos: that is why 0xffffffff
<marex> apalos: you might need to fix up the whole ulong passing around and make it u64 I think
<steffen[m]> marex: thanks, yes that might help. I'll see if we can do something similar
<apalos> oh 'great' ram_top is unsigned long
naoki has quit [Quit: naoki]
<apalos> marex: or tweak the lmb code to expect the max address on +1
<marex> apalos: yes, the legacy guts
<apalos> but since LPAE is real we might be better of to doing a u64 to begin with
<apalos> bleh
<marex> apalos: but ... contemporary 32bit ARM will suffer performance penalty and size growth
<apalos> marex: that imx platform is 64bit though no ?
<marex> apalos: which imx platform ?
<marex> apalos: imx1..imx7d is 32bit
<apalos> imx8m
<marex> apalos: imx8 and onward is 64bit
<apalos> yea so the commit is for imx8m
<apalos> We shgould be ok setting it to 0x100000000
<marex> apalos: board_get_usable_ram_top returns ulong
<marex> apalos: isn't that still 32bit on arm64 ?
<marex> let me get my decoder ring
<apalos> hrmm good question dont remember, let me check
<marex> minimum 32, means it can be 32 or 64bit
<marex> if we want to be surely 64bit, that is u64 or unsigned long long
<apalos> nop it's 8
<apalos> so we are fine for now
<apalos> but yes I agree that ulong needs to die
<marex> apalos: that's like an embedded footbazooka right there
<apalos> marex: yes hence the fixed width characters on teh c spec :>
<marex> apalos: so the question is, what is the simplest fix for 2025.01 ? patch LMB ? and in the long run, plumb in u64 ?
<apalos> patch that board code and in the long run fix the u64 AFAICT
<apalos> because that ulong is in phys_addr_t as well
<apalos> So you would have bigger problems if that was 32bits
<apalos> because if we change LMB we need to go back and make the boards that do it correctly already do +1
<apalos> which is riskier imho
<apalos> but yes looking a bit more, for arm64 at least this is risky regardless
<apalos> phys_addr_t is unsigned long long....
<apalos> and the ulong is unsigned long and we hope that stars will align
<apalos> \o/
<apalos> but I still think it's fine for the imx8m stuff
mmu_man has joined #u-boot
<marex> apalos: I have > 4 GiB MX8MP here if needed
goliath has quit [Quit: SIGSEGV]
warpme has quit [Quit: Textual IRC Client: www.textualapp.com]
warpme has joined #u-boot
schmendrick3312 has joined #u-boot
schmendrick67 has joined #u-boot
schmendrick67 has quit [Client Quit]
schmendrick3312 has quit [Ping timeout: 240 seconds]
<apalos> marex: thanks, I'll have a look later tonight
<apalos> because, yes we can define it as such in 64bits but....
<apalos> We will cretae a rule that says "yea on 64 bits it's <usable> + 1, but on 32bit it's <usable>"
<apalos> which is a horrible design
<apalos> otoh on 32bit boards LPAE (or any board that doesnt start at 0x0) will have other problems
<pivi> A long time ago in a galaxy far, far away .... you could go to the moon with just 8-bit ... :-)
frieder has quit [Remote host closed the connection]
mmu_man has quit [Ping timeout: 252 seconds]
joeskb7_ has joined #u-boot
joeskb7 has quit [Ping timeout: 248 seconds]
ldevulder has quit [Quit: Leaving]
warpme has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
mckoan is now known as mckoan|away
flyback has quit [Remote host closed the connection]
joeskb7 has joined #u-boot
joeskb7_ has quit [Ping timeout: 265 seconds]
flyback has joined #u-boot
sszy has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.]
persmule_ has joined #u-boot
persmule has quit [Ping timeout: 264 seconds]
mripard has quit [Quit: WeeChat 4.4.4]
haritz has joined #u-boot
haritz has joined #u-boot
haritz has quit [Changing host]
vagrantc has joined #u-boot
joeskb7 has quit [Quit: Lost terminal]
<Tartarus> shadow: around? I want to talk about that riscv platform and testing it
ikarso has quit [Quit: Connection closed for inactivity]
joeskb7 has joined #u-boot
joeskb7 has quit [Quit: Lost terminal]
joeskb7 has joined #u-boot
dujem has joined #u-boot
mmu_man has joined #u-boot
rvalue is now known as WastedGymMembers
WastedGymMembers is now known as DntBuyGymPlan
mmu_man has quit [Ping timeout: 265 seconds]
mmu_man has joined #u-boot
DntBuyGymPlan has quit [Quit: ChristmasMaintenance]
slobodan__ has quit [Read error: Connection reset by peer]
slobodan has joined #u-boot
goliath has joined #u-boot
flyback has quit [Ping timeout: 252 seconds]
slobodan_ has joined #u-boot
slobodan has quit [Ping timeout: 252 seconds]
flyback has joined #u-boot
slobodan has joined #u-boot
slobodan_ has quit [Ping timeout: 244 seconds]
persmule_ has quit [Ping timeout: 264 seconds]
gsz has quit [Ping timeout: 244 seconds]
persmule_ has joined #u-boot
slobodan has quit [Read error: Connection reset by peer]
slobodan has joined #u-boot
slobodan_ has joined #u-boot
mmu_man has quit [Ping timeout: 244 seconds]
ikarso has joined #u-boot
slobodan has quit [Ping timeout: 244 seconds]
jfsimon1981 has quit [Ping timeout: 276 seconds]
mmu_man has joined #u-boot
lehmanju has quit [Quit: The Lounge - https://thelounge.chat]
lehmanju has joined #u-boot
jfsimon1981 has joined #u-boot
prabhakalad has quit [Quit: Konversation terminated!]
prabhakalad has joined #u-boot
jfsimon1981 has quit [Ping timeout: 248 seconds]
jfsimon1981 has joined #u-boot
jfsimon1981 has quit [Remote host closed the connection]
jfsimon1981 has joined #u-boot