not freeing initmem?
There is still one or more __init 's too much somewhere. If we would free the memory without fixing/finding this place first, the kernel will crash.
Helge
Quoted 22 lines · Ralf Hildebrandt
From my dmesg output of the recent 2.4.18-pa3:
...
EXT3-fs: mounted filesystem with ordered data mode.
VFS: Mounted root (ext3 filesystem) readonly.
NOT FREEING INITMEM (351k)
Adding Swap: 124800k swap-space (priority -1)
EXT3 FS 2.4-0.9.17, 10 Jan 2002 onsd(8,3), internal journal
...What does that mean? Why is the memory not freed on PA-RISC, but freed on x86?
On my x86 box:
...
EXT3-fs: mounted filesystem with ordered data mode.
VFS: Mounted root (ext3 filesystem) readonly. Freeing unused kernel memory: 224k freed
EXT3 FS 2.4-0.9.17, 10 Jan 2002 onide0(3,2), internal journal Real Time Clock Driver v1.10e
...
Signature · 4 lines
Helge Deller SAP AG - LinuxLab and Unix Platforms email: [email protected] Phone: +49 6227 7 - 67783
Quoted 3 lines
There is still one or more
__init's too much somewhere. If we would free the memory without fixing/finding this place first, the kernel will crash.
Disable interrupts, memset all the freed pages to an invalid instruction in parisc asm at the point you free the init pages, then re-enable ints and the traceback should give you a direct stack trace to the offender
Am 06.03.2002 um 16:19:13 +0100 schrieb Ralf Hildebrandt folgendes:
Quoted 2 lines
What does that mean? Why is the memory not freed on PA-RISC, but freed on x86?
OK, I grepped for that string and found the infamous FIXME:
/* FIXME: */
printk(KERN_INFO "NOT FREEING INITMEM (%dk)\n",
(&__init_end - &__init_begin) >> 10);
return; OTOH, why does the code below that not work for PA-RISC?