<sjg1>
marex: I would like that patch a lot more if it added a test
rbox has joined #u-boot
[[R]] has quit [Ping timeout: 255 seconds]
rbox is now known as [[R]]
mmu_man has quit [Ping timeout: 250 seconds]
Slimey has quit [Remote host closed the connection]
mwalle has quit [Quit: WeeChat 2.3]
Peng_Fan has quit [Quit: Connection closed for inactivity]
fdanis_away is now known as fdanis
frieder has joined #u-boot
mckoan|away is now known as mckoan
stefanro has quit [Remote host closed the connection]
stefanro has joined #u-boot
Peng_Fan has joined #u-boot
mranostaj has quit [Remote host closed the connection]
mwalle has joined #u-boot
matthias_bgg has joined #u-boot
smartin has joined #u-boot
mmu_man has joined #u-boot
monstr has joined #u-boot
<marex>
sjg1: last time a patch was rejected with the argument "I would like it a lot more if it had a test" , it popped up a year later with a CVE tag, so let me be clear I am really starting to be allergic to this "test" comment
jwillikers has joined #u-boot
torez has joined #u-boot
jwillikers has quit [Remote host closed the connection]
jwillikers has joined #u-boot
hanetzer has quit [Ping timeout: 265 seconds]
hanetzer has joined #u-boot
macromorgan has quit [Killed (zirconium.libera.chat (Nickname regained by services))]
macromorgan has joined #u-boot
smartin1 has joined #u-boot
pbrobinson_ has joined #u-boot
swiftgeek_ has joined #u-boot
fdanis` has joined #u-boot
pbrobinson has quit [Ping timeout: 255 seconds]
swiftgeek has quit [Ping timeout: 255 seconds]
smartin has quit [Ping timeout: 255 seconds]
elafon has quit [Ping timeout: 255 seconds]
apalos has quit [Ping timeout: 255 seconds]
fdanis has quit [Ping timeout: 255 seconds]
pbrobinson_ is now known as pbrobinson
smartin1 is now known as smartin
apalos has joined #u-boot
elafon has joined #u-boot
hanetzer has joined #u-boot
hanetzer has quit [Changing host]
<sjg1>
marex: We just need to write tests. You often complain about regressions and test are how you stop them
<sjg1>
apalos: "changing page tables etc, needs a lot of work, there's currently limited code into u-boot"
<sjg1>
apalos: So if you actually want to protect memory, that is the feature you need to add
<sjg1>
apalos: There is some ARM code there, but it needs a proper API so it can be supported on any arch (when people add the impl)
<apalos>
sjg1: even if that's the case the DTB will *still* need to be on r/w memory
<apalos>
because we do fixups and xypron has proposed a protocol that efi apps (i.e grub) can apply fixups as well
<apalos>
so I dont get the 'protect the dtb point'
<apalos>
as much as I dont get why we like ignoring the dt spec
<apalos>
'the aim is align the dtb, but wait let's put a bunch of u-boot only stuff there'
<apalos>
that's what the argument feels like
<mrnuke>
sjg1: with the risk of being off-topic, I couldn't tell if you were speaking british or australian
akaWolf has quit [Ping timeout: 240 seconds]
<apalos>
~.~.~.
<apalos>
oops, not ssh
jwillikers has quit [Remote host closed the connection]
<apalos>
sjg1: but we can ingore the DT spec for all I care. It's a separate DT we need to crete anyway
<apalos>
but if we want to what you say is *demand* the DT is configured as CONFIG_OF_BOARD and then we fix the mmu and switch those pages to /ro
<apalos>
that would get us the same result
Ododo has joined #u-boot
<marex>
sjg1: your method of motivating me to write tests has the opposite effect, that's all
LeSpocky has joined #u-boot
<LeSpocky>
hello
<LeSpocky>
I'm currently working on support for a new board with a sama5d27 d5m soc … I get a u-boot shell and can access the sd card so far
<LeSpocky>
now I switched on some more options and now it does not boot to the shell prompt anymore but hangs when relocating
<LeSpocky>
last debug messages (two lines):
<LeSpocky>
Relocating to 23f61000, new gd at 23b20ea0, sp at 23b1dae0
<LeSpocky>
initcall: 23f11c44
<LeSpocky>
what might be the problem here?
<LeSpocky>
u-boot image is 603K and soc has 64 MiB ram
<wyre>
how can I generate .scr files?
<wyre>
I mean, I'm curious because apparently I can give these scripts to U-Boot from a tftp to be run
<marex>
wyre: mkimage -T script
<marex>
LeSpocky: is it relocating to valid DRAM location ?
<LeSpocky>
last message now is 'Pre-reloc malloc() used 0x1024 bytes (4 KB)' from function initr_malloc()
<marex>
yeah
<marex>
so malloc init fails ?
<marex>
just keep adding printf()s into the initr_malloc()
<LeSpocky>
^^
<marex>
is your early malloc area even pointing to valid memory ?
<LeSpocky>
there's one message 'Reserving 2304k for malloc() at: 23d4a000' before relocation, not sure if that's the early or the later address, but that's inside the 64 MiB memory, yes
<Ododo>
(i have this tpm device on my chip i don't know what exactly what i'm gonna do with it)
<wyre>
marex, what about the tftp alias? 🤔
<LeSpocky>
'using memory 0x23b49000-0x23f89000 for malloc()' looks good, I have no idea :-/
<LeSpocky>
i put it aside now, will continue after my holidays
mckoan is now known as mckoan|away
<rfried>
marex, wayre: Interesting, I'm not sure, I need to look it up.
<milkylainen_>
Tartarus: I still can't get bloody stdin from usbkbd to work no matter what I try. :\
<milkylainen_>
stdout = vidconsole (efi framebuffer), stdin = usbkbd. xhci, ehci, uhci compiled in with pci variants. kbd support enabled, tried different kbd polling methods. disabled 16550 uart due to bad probe and usage later (hang)... still nothing.
<milkylainen_>
usb start was added because I was missing the preboot variable but that does not seem to change anything.
<milkylainen_>
I can see the usb probe (over xhci) finding 5-7 devices depending on whats plugged.
monstr has quit [Remote host closed the connection]
<LeSpocky>
c ya
mranostaj has quit [Remote host closed the connection]
mranostaj has joined #u-boot
<cambrian_invader>
Ododo: typically, the public key is just included with U-Boot's devicetree
<cambrian_invader>
because you also need to verify U-Boot itself (since if you allow arbitrary code to run it can just load whatever it wants)
frieder has quit [Remote host closed the connection]
<sjg1>
marex: What do you mean? What is my method?
<sjg1>
marex: When making a change or adding a feature, I think the first think people should consider is how to write the test
<marex>
sjg1: well, then that's good and all, so if you want to write tests, go right ahead and write tests
<sjg1>
marex: "your method of motivating me to write tests has the opposite effect, that's all"
<sjg1>
marex: I just don't know what you are referring to, or what you are asking for
<marex>
the constant nagging with "does it have a test", which results to bugfixes being missed and even later popping up again as CVEs
<marex>
sjg1: if you want people to write tests, then help out by writing those tests for a few submitted patches, maybe someone will pick up on that
<marex>
but just telling people "yeah, I would like your patch more if it had a test" is really not getting too far, esp. if the documentation for the test framework is about as hard to digest as the test framework itself, which is outright cryptic
swiftgeek_ is now known as swiftgeek
<sjg1>
marex: Lots of people have figured this out. Perhaps you could spend a few hours trying it
<sjg1>
marex: As to writing tests for submitted patches, take a look at str_ut.c
<sjg1>
or print_ut.c
<sjg1>
marex: The idea that nagging about writing a test (which is just as frustrating for me) causes missed bugs seems odd to me
<sjg1>
marex: Anyway thanks for explaining how you feel. I hope we now understand each other\
<marex>
sjg1: yes, this is pretty much it
<marex>
if you start dropping bugfix patches because they dont have a test, then you will end up with bugs in the codebase which could have been fixed
smartin has quit [Remote host closed the connection]
<cambrian_invader>
but if you don't include a test what is to stop the same bug from being reintroduced?
vagrantc has joined #u-boot
birkoff has quit [Ping timeout: 255 seconds]
torez has quit [Quit: torez]
vagrantc has quit [Quit: leaving]
birkoff has joined #u-boot
birkoff has joined #u-boot
birkoff has quit [Changing host]
LeSpocky has quit [Quit: reboot]
matthias_bgg has quit [Quit: Leaving]
jwillikers has joined #u-boot
guillaume_g has quit [Ping timeout: 255 seconds]
sbach has joined #u-boot
sbach has quit [Remote host closed the connection]