Affected kernel : 2.0.36 Bug description : When using 'append="mem=128M" ' in 'lilo.conf', kernel will panic Affected system : ASUS P5A Super 7 Mainboard Disk controller : Adaptec 2940U controller Hard drives : IBM 4.3G and Quantum 4.3G System memory : 128Mb CPU : AMD K6-2 350Mhz Bug causes "Kernel panic" when passing information on amount of memory available from LILO to Kernel. To reproduce bug do the following: In 'lilo.conf' add this line append="mem=128M" When doing this the following result is reported before system hangs: ----0000---- Initrd extends beyond end of memory (0x00fff190>0x00000080) disabling initrd ...some other unrelated messages... Memory 0K/0K Available (0K kernel code, 0K reserved, 0K data) Kernel panic: Failed to allocate buffer hash table -----0000---- This does not happen if using 'append="mem=96M"' nor does it happen when using kernel 2.0.34 Bug discovered during upgrade of motherboard memory to 128Mb and not realizing that kernel 2.0.36 should detect memory > 64Mb automatically. Removing the 'append' line fixes the problem. Kernel should not "zero" out the memory detected when the same amount is passed from LILO. It seems that 2 digit memory numbers do not cause the problem, only 3 digit. Submitted by Anthony White, Australia E-Mail: awhite.net.au or anthony_w
I tried this problem on two different ASUS motherboards with 128 megs of ram withno problem .
From BootPrompt-HOWTO : ``The kernel will accept any `mem=xx' parameter you give it, and if it turns out that you lied to it, it will crash horribly sooner or later. [skipped - AN] NOTE NOTE NOTE: some machines might use the top of memory for BIOS cacheing or whatever, so you might not actually have up to the full xxMB addressable.