1 |
Ok, the other thread really took off down some odd, and to my eyes way off |
2 |
topic, directions. |
3 |
|
4 |
Anyway, the idea was to ask for feedback on who has done it and the results |
5 |
they got. To keep things focussed, lets leave out any talk of ARM, cross |
6 |
compiling, small/embedded systems, iphones, Apple, Android, old-fartiness |
7 |
credentials etc :) |
8 |
|
9 |
For context, my system is a 4 core AMD with 10gig ram, SSD for most of the |
10 |
OS and VMs, then software RAID1 for bulk storage. |
11 |
|
12 |
Since yesterday, uptime = 20.27 and no issues noticed. |
13 |
|
14 |
<original email> |
15 |
The CFLAGs used when building the kernel are set in the Makefile; |
16 |
# grep HOSTCFLAGS /usr/src/linux/Makefile |
17 |
HOSTCFLAGS = -Wall -Wmissing-prototypes -Wstrict-prototypes -O2 |
18 |
-fomit-frame-pointer |
19 |
|
20 |
To build with other flags you set CFLAGS_KERNEL, so i've added a suitable |
21 |
-march to the standard ones for my system; |
22 |
export CFLAGS_KERNEL=" -Wall -Wmissing-prototypes -Wstrict-prototypes -O2 |
23 |
-fomit-frame-pointer -pipe -march=amdfam10" |
24 |
|
25 |
then make, install, reboot. In my case the riced kernel is the same size as |
26 |
the regular one, but the md5 is different. Its been up for an hour with no |
27 |
obvious issues, and does seem snappier, but of course my brain is full of |
28 |
cognitive bias. Has anyone else played with this? Any good or bad outcomes? |
29 |
</original email> |