Alan Cox wrote:
>
> > 2.2.18-pre with good results on Thinkpad 600x. No adverse consequencies
> > (memory is reported correctly, suspend/hibernate work fine). I also had
> > a chance to use it on several big SMP and UP servers (where e801 memory
> > reporting works fine) Intel, Compaqs and IBM. No adverse consequencies
> > either. What do you say?
>
> Until every 2.4.0 doesnt boot case has been accounted for (within reason anyway)
> no. Wrong mem sizing returns that are not caught are a great way to take out
> your entire root fs
Funny, my impression from reading the lk is that boot failures are not
connected with E820 memory detection. It is (at least in the backport
patch) configurable and can be turned off. Have people who experience
boot failures tried booting with E820 disabled? I know there were some
problems with E820 on some machines but Linus has fixed it himself in
early 2.4.x.
As for corruption, manually specifying mem=xxx to the kernel is no less
dangerous. Probably more, because subtle problems are possible. Consider
all those cases when people use mem=128m when they sould have used
mem=127m because of BIOS, ACPI, video, or whatever. What happens then? A
crash and a serious root fs corruption 10 hours after boot? At least
with E820 (if we do not expect to work for all machines) people will see
at boot time that memory was misdetected.
Of course, it is just a MHO. I am just greatly annoyed that most laptops
and home computers these days will report 64M of RAM if e801 scheme is
used. Servers do not have this problem. It is too late now to fight that
standard (PC 99). We have to embrace it now, otherwise it causes
slanderers to say that Linux does not support above 64M of RAM. Again,
this just a MHO.
-- ---------------------------------------- Constantine Gavrilov Unix System Administrator and Programmer Xpert Integrated Systems 1 Shenkar St, Herzliya 46725, Israel Phone: (972-8)-952-2361 Fax: (972-9)-952-2366 ---------------------------------------- - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Mon Oct 23 2000 - 21:00:09 EST