1 commit 6c59dbf0e13ca7d36842059afebc794df39feb7e
2 Author: Oliver Pinter <oliver.pntr@gmail.com>
3 Date: Tue Feb 26 17:06:12 2008 +0100
7 Signed-off-by: Oliver Pinter <oliver.pntr@gmail.com>
9 commit 65546a80c00aa009f5b3295f217c350a31f19e63
10 Author: David Howells <dhowells@redhat.com>
11 Date: Thu Feb 21 16:12:45 2008 +0000
13 MM: Fix macro argument substitution in PageHead() and PageTail()
15 Fix macro argument substitution in PageHead() and PageTail() - 'page' should
16 have brackets surrounding it (commit 6d7779538f765963ced45a3fa4bed7ba8d2c277d).
18 Signed-off-by: David Howells <dhowells@redhat.com>
19 Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
20 Signed-off-by: Oliver Pinter <oliver.pntr@gmail.com>
22 commit b98ecf77bf36a255ad82c2320c3394ea762e840d
23 Author: Kurt Garloff <garloff@suse.de>
24 Date: Mon Feb 18 21:16:27 2008 +0100
26 make (low) swappiness safer to use
29 make swappiness safer to use
30 has been added to the -mm tree. Its filename is
31 make-swappiness-safer-to-use.patch
33 See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
34 out what to do about this
36 #Subject: make swappiness safer to use
37 #From: Andrea Arcangeli <andrea@suse.de>
39 Swappiness isn't a safe sysctl. Setting it to 0 for example can hang a
40 system. That's a corner case but even setting it to 10 or lower can waste
41 enormous amounts of cpu without making much progress. We've customers who
42 wants to use swappiness but they can't because of the current
43 implementation (if you change it so the system stops swapping it really
44 stops swapping and nothing works sane anymore if you really had to swap
45 something to make progress).
47 This patch from Kurt Garloff makes swappiness safer to use (no more huge
48 cpu usage or hangs with low swappiness values).
50 I think the prev_priority can also be nuked since it wastes 4 bytes per
51 zone (that would be an incremental patch but I wait the nr_scan_[in]active
52 to be nuked first for similar reasons). Clearly somebody at some point
53 noticed how broken that thing was and they had to add min(priority,
54 prev_priority) to give it some reliability, but they didn't go the last
55 mile to nuke prev_priority too. Calculating distress only in function of
56 not-racy priority is correct and sure more than enough without having to
57 add randomness into the equation.
59 Patch is tested on older kernels but it compiles and it's quite simple
62 Overall I'm not very satisified by the swappiness tweak, since it doesn't
63 rally do anything with the dirty pagecache that may be inactive. We need
64 another kind of tweak that controls the inactive scan and tunes the
65 can_writepage feature (not yet in mainline despite having submitted it a
66 few times), not only the active one. That new tweak will tell the kernel
67 how hard to scan the inactive list for pure clean pagecache (something the
68 mainline kernel isn't capable of yet). We already have that feature
69 working in all our enterprise kernels with the default reasonable tune, or
70 they can't even run a readonly backup with tar without triggering huge
71 write I/O. I think it should be available also in mainline later.
73 Cc: Nick Piggin <npiggin@suse.de>
74 Signed-off-by: Kurt Garloff <garloff@suse.de>
75 Signed-off-by: Andrea Arcangeli <andrea@suse.de>
76 Signed-off-by: Fengguang Wu <wfg@mail.ustc.edu.cn>
77 Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
78 Signed-off-by: Oliver Pinter <oliver.pntr@gmail.com>
80 commit 853c342ffcde7ca697f4b7950aa2a12f1752fce1
81 Author: Thomas Gleixner <tglx@linutronix.de>
82 Date: Wed Feb 20 00:29:02 2008 +0100
84 genirq: do not leave interupts enabled on free_irq
86 commit 89d694b9dbe769ca1004e01db0ca43964806a611
88 The default_disable() function was changed in commit:
90 76d2160147f43f982dfe881404cfde9fd0a9da21
91 genirq: do not mask interrupts by default
93 It removed the mask function in favour of the default delayed
94 interrupt disabling. Unfortunately this also broke the shutdown in
95 free_irq() when the last handler is removed from the interrupt for
96 those architectures which rely on the default implementations. Now we
97 can end up with a enabled interrupt line after the last handler was
98 removed, which can result in spurious interrupts.
100 Fix this by adding a default_shutdown function, which is only
101 installed, when the irqchip implementation does provide neither a
102 shutdown nor a disable function.
104 Pointed-out-by: Michael Hennerich <Michael.Hennerich@analog.com>
105 Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
106 Acked-by: Ingo Molnar <mingo@elte.hu>
107 Tested-by: Michael Hennerich <Michael.Hennerich@analog.com>
108 Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
109 Signed-off-by: Oliver Pinter <oliver.pntr@gmail.com>
111 commit 55f2876a0023596d6b27e20b0c806654b38ccfae
112 Author: Ingo Molnar <mingo@elte.hu>
113 Date: Fri Feb 15 20:58:22 2008 +0100
115 x86_64: CPA, fix cache attribute inconsistency bug, v2.6.22 backport
117 fix CPA cache attribute bug in v2.6.2[234]. When phys_base is nonzero
118 (when CONFIG_RELOCATABLE=y) then change_page_attr_addr() miscalculates
119 the secondary alias address by -14 MB (depending on the configured
122 The default 64-bit kernels of Fedora and Ubuntu are affected:
124 $ grep RELOCA /boot/config-2.6.23.9-85.fc8
127 $ grep RELOC /boot/config-2.6.22-14-generic
130 and probably on many other distros as well.
132 the bug affects all pages in the first 40 MB of physical RAM that
133 are allocated by some subsystem that does ioremap_nocache() on them:
135 if (__pa(address) < KERNEL_TEXT_SIZE) {
137 Hence we might leave page table entries with inconsistent cache
138 attributes around (pages mapped at both UnCacheable and Write-Back),
139 and we can also set the wrong kernel text pages to UnCacheable.
141 the effects of this bug can be random slowdowns and other misbehavior.
142 If for example AGP allocates its aperture pages into the first 40 MB
143 of physical RAM, then the -14 MB bug might mark random kernel texto
144 pages as uncacheable, slowing down a random portion of the 64-bit
145 kernel until the AGP driver is unloaded.
147 Signed-off-by: Ingo Molnar <mingo@elte.hu>
148 Acked-by: Thomas Gleixner <tglx@linutronix.de>
149 Signed-off-by: Oliver Pinter <oliver.pntr@gmail.com>