IRC channel logs

2024-06-20.log

back to list of logs

<oriansj>cosinusoidally: well yes, cramming a 13MB bzImage into a 1.44MB floppy was not really an option
<oriansj>in Linux 2.6.9 36% of Drivers were just PCI; 35% USB; 9% GPU; 8.5% SCSI; 2% IDE. So taking out PCI and USB drivers that you don't need would be a serious size reduction
<oriansj>and I just realized that I forgot to include the unxz and wrap programs in stage0-posix's readme
<oriansj>better fix that
<matrix_bridge><Andrius Štikonas> oriansj: also note that unxz is miscompiled by M2-Planet on riscv
<aggi>cosinusoidally: i am not opposed to RISC-V or ARM, for various reasons, such as the firmware situation is a little less hostile with some of them
<aggi>and some devices are ultra-low power consumption
<aggi>however, with regards to bootstrapping, these platforms got no path to avoid x86 as a buildhost, require additional cross-compilation steps
<aggi>and finally do not solve problems there is with bootstrapping
<aggi>it is merely _additional_ efforts required with RISC-V and ARM, which has to be justified by those who seek support for them
<aggi>u-boot loader and device-trees didn't arrive at providing a generic interface for generic kernel support that is/was possible with x86
<aggi>even worse, u-boot introduced a transitive dependency towards c++, for the bootloader, which is a hazard for bootstrapping
<aggi>this is where i see many bootstrapping problems rooted: http://web.archive.org/web/20210122075331/https://www.nytimes.com/2002/05/12/us/for-parts-nasa-boldly-goes-on-ebay.html
<aggi>in practice, there is no way around x86 for bootstrapping, and this platform is rigged
<aggi>recently, another historical note: Zilog announced they will terminate Z80 production too
<oriansj>stikonas: yeah we probably need to debug the unxz behavior on RISC-V vs x86/amd64 and figure out the why for the difference in behavior.
<oriansj>aggi: shouldn't be an issue as the z80 has been completely reverse engineered: http://www.visual6502.org/JSSim/expert-z80.html
<oriansj>so we could with a little money start making our own z80 chips (same for 6502 chips)
<oriansj>and ultimately we are going to have to make our own hardware to rule out a Nexus Intruder Program class attacker infecting our Lithography process.
<aggi>i'll depart for a few weeks