<NishanthMenon>
bryanb: oops.. wrong thread.. my bad. please address my comments.
<bryanb>
NishanthMenon: sweet! I can do that :)
camus has joined #u-boot
d-s-e has quit [Ping timeout: 240 seconds]
mncheck has joined #u-boot
mmu_man has quit [Ping timeout: 246 seconds]
mmu_man has joined #u-boot
stefanro has quit [Quit: Leaving.]
FergusL has quit [Read error: Connection reset by peer]
FergusL has joined #u-boot
<rfs613>
marex: how is the grey matter doing today?
___nick___ has joined #u-boot
<marex>
rfs613: its dead, Jim
<marex>
rfs613: what's up?
<rfs613>
marex: well you better get Bones to take a look at it then ;-)
<rfs613>
marex: about the last three lines in arch/arm/mach-rmobile/Makefile
d-s-e has joined #u-boot
d-s-e has quit [Client Quit]
<rfs613>
it looks like you added them around 2019.07
<rfs613>
sjg1 made a change in 2020.10 but it's just renaming of variables
* marex
looks
<marex>
rfs613: lemme set up a build of something first
<rfs613>
ok
<marex>
oh the sreccat goo
<marex>
rfs613: so, what about it ?
<rfs613>
the intention of those lines is to check if srec_cat is available, and if so, to build the *.scif format.
<rfs613>
and the logic is perfectly fine, however, it turns out that INPUTS-y already contains the *.scif format
<rfs613>
so the check is ineffective, it will call srec_cat regardless... and if srec_cat is not installed, the build will fail.
<rfs613>
you can see it for example building for eagle board.
goliath has quit [Quit: SIGSEGV]
<marex>
rfs613: if you could figure out how to not do any .scif file and rather assemble the same with just compiler+linker, that would be great
mmu_man has quit [Ping timeout: 264 seconds]
<rfs613>
marex: part of the issue is CONFIG_SPL_TARGET in common/spl/Kconfig, which defaults to enable .scif for RCAR_GEN3
<rfs613>
that in turn leads to it being added to INPUT-y in the top-level Makefile
<rfs613>
I agree that using assembler would be nice instead of inserting the binary. As for using the linker to generate the image, I can take a stab at it, but I cannot test (don't have hardware) and suspect i will not end up with a binary that is identical to the .scif
<rfs613>
out of the the ~20 renesas boards which have defconfigs in the tree, this affects all but 4 of them, btw.
<marex>
if you end up with the same .srec file, then great
<rfs613>
it seems the .scif depends on the .srec, or am I misreading things?
<marex>
btw I dont see any INPUT-y anywhere
<marex>
it does, the .scif is just srec+header
<rfs613>
marex: it is line 993 in the top-level Makefile that adds the .scif to INPUT-y
<marex>
Hypfer: in general, no, but the hush 2021 update might bring support for it
<marex>
Hypfer: what is your use case exactly ?
<Hypfer>
marex aah okay that explains that. I was a bit confused as the https://hush-shell.github.io/cmd/capture.html docs do state that it should be possible but I wasn't sure if that is even the same hush shell
<Hypfer>
just curiosity basically. I'm quite impressed by what is possible with u-boot
<marex>
Hypfer: no, pipes and redirections are not implemented by the U-Boot built-in hush fork
<marex>
Hypfer: but there are patches posted to update the hush to newer one