Originally Posted by asdfblah
Actually, i've had one issue too with kernels 3.15+ which might be related? Since kernel 3.0 i've been successfully building on a customised .config file without an initrd image (initial ramdisk options disabled, debugging completely pulled out etc bare essentials) but i couldn't get 3.15+ to boot originally (because a module was being overlooked).
When 3.14 became the new latest 'longterm' version, i stuck with it since it runs without issue, but when i came to try 3.16+, i noticed that i could only build a bootable image if i edited the Kconfig file in the arch/x86 directory and made the config IOSF_MBI entry default to n instead of m - and even then, the kernel does say it's searching for a ramdisk on startup with these later kernels, even though i've always expressly removed support for it and it's never been an issue before? This issue has been present in every kernel, rc or final since 3.15-rc1?
(I've messed around with so much stuff over the years, this issue may very well be my own fault ), but i thought i'd bring it up on the off chance it's not and it's actually relevant. Changing the "IOSF_MBI" to default to 'y' might make a difference and include the functionality if it's suddenly become essential.
Just looking at the Kconfig file from the 3.14 kernel, it reads:
whereas from kernel 3.15 onwards, it now reads:
depends on PCI
To be selected by modules requiring access to the Intel OnChip System
Fabric (IOSF) Sideband MailBox Interface (MBI). For MBI platforms
enumerable by PCI.
I've got a Radeon HD 4200, "Turion II Dual-Core Mobile M52" CPU, no such _OSC option in my BIOS(?), and with the exception of having to make this modification to get my initrd-less kernel bootable, and it mysteriously looking for a ramdisk for a fraction of a second at startup, everything is running well.
depends on PCI