On July 26, 2020 11:24:25 PM PDT, Christoph Hellwig <hch@xxxxxx> wrote: >On Sun, Jul 26, 2020 at 11:20:41PM -0700, hpa@xxxxxxxxx wrote: >> On July 26, 2020 8:05:34 PM PDT, Al Viro <viro@xxxxxxxxxxxxxxxxxx> >wrote: >> >On Tue, Jul 14, 2020 at 09:04:22PM +0200, Christoph Hellwig wrote: >> >> Don't rely on the implicit set_fs(KERNEL_DS) for ksys_open to >work, >> >but >> >> instead open a struct file for /dev/console and then install it as >FD >> >> 0/1/2 manually. >> > >> >I really hate that one. Every time we exposed the internal details >to >> >the fucking early init code, we paid for that afterwards. And this >> >goes over the top wrt the level of details being exposed. >> > >> >_IF_ you want to keep that thing, move it to fs/file.c, with dire >> >comment >> >re that being very special shite for init and likely cause of >> >subsequent >> >trouble whenever anything gets changed, a gnat farts somewhere, etc. >> > >> > Do not leave that kind of crap sitting around init/*.c; KERNEL_DS >> >may be a source of occasional PITA, but here you are trading it for >a >> >lot >> >worse one in the future. >> >> Okay... here is a perhaps idiotic idea... even if we don't want to >run stuff in actual user space, could we map initramfs into user space >memory before running init (execing init will tear down those mappings >anyway) so that we don't need KERNEL_DS at least? > >Err, why? The changes have been pretty simple, and I'd rather not come >up with new crazy ways just to make things complicated. Why? To avoid this neverending avalanche of special interfaces and layering violations. Neatly deals with non-contiguous contents and initramfs in device memory, etc. etc. etc. -- Sent from my Android device with K-9 Mail. Please excuse my brevity.
