| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <20061113141455.GA14982@elte.hu>
Date: Mon, 13 Nov 2006 15:14:56 +0100
From: Ingo Molnar <mingo@...e.hu>
To: Andi Kleen <ak@...e.de>
Cc: "Siddha, Suresh B" <suresh.b.siddha@...el.com>,
Andrew Morton <akpm@...l.org>, linux-kernel@...r.kernel.org,
ashok.raj@...el.com
Subject: Re: [patch] genapic: optimize & fix APIC mode setup
* Andi Kleen <ak@...e.de> wrote:
> > Also, i'd like to have a description of how to reproduce those CPU
> > hotplug problems, so that i can try to fix it.
>
> iirc they just did stress tests. Plug/unplug cpus in a tight loop and
> do some workloads and see what happens.
i just tested it and it works fine for me.
Andrew, please apply my patch, which fixes a real regression on my box,
until some detailed instruction / testcase is shown about how to
reproduce whatever supposed CPU-hotplug bug is hidden by this kludge -
at which point i offer to fix that CPU-hotplug bug too. (whatever it
takes)
(I cannot believe that this kludge was allowed into the x86_64 tree and
that its non-removal is now justified with 'it breaks something around
the CPU hotplug code, go figure it out yourself'. That's a sure way to
let the kernel bitrot...)
Ingo
------------------------>
Subject: [patch] genapic: optimize & fix APIC mode setup
From: Ingo Molnar <mingo@...e.hu>
this patch fixes a couple of inconsistencies/problems i found while
reviewing the x86_64 genapic code (when i was chasing mysterious eth0
timeouts that would only trigger if CPU_HOTPLUG is enabled):
- AMD systems defaulted to the slower flat-physical mode instead
of the flat-logical mode. The only restriction on AMD systems
is that they should not use clustered APIC mode.
- removed the CPU hotplug hacks, switching the default for small
systems back from phys-flat to logical-flat. The switching to logical
flat mode on small systems fixed sporadic ethernet driver timeouts i
was getting on a dual-core Athlon64 system:
NETDEV WATCHDOG: eth0: transmit timed out
eth0: Transmit timeout, status 0c 0005 c07f media 80.
eth0: Tx queue start entry 32 dirty entry 28.
eth0: Tx descriptor 0 is 0008a04a. (queue head)
eth0: Tx descriptor 1 is 0008a04a.
eth0: Tx descriptor 2 is 0008a04a.
eth0: Tx descriptor 3 is 0008a04a.
eth0: link up, 100Mbps, full-duplex, lpa 0xC5E1
- The use of '<= 8' was a bug by itself (the valid APIC ids
for logical flat mode go from 0 to 7, not 0 to 8). The new logic
is to use logical flat mode on both AMD and Intel systems, and
to only switch to physical mode when logical mode cannot be used.
If CPU hotplug is racy wrt. APIC shutdown then CPU hotplug needs
fixing, not the whole IRQ system be made inconsistent and slowed
down.
- minor cleanups: simplified some code constructs
build & booted on a couple of AMD and Intel SMP systems.
Signed-off-by: Ingo Molnar <mingo@...e.hu>
---
arch/x86_64/kernel/genapic.c | 54 ++++++++++++++++---------------------------
1 file changed, 21 insertions(+), 33 deletions(-)
Index: linux/arch/x86_64/kernel/genapic.c
===================================================================
--- linux.orig/arch/x86_64/kernel/genapic.c
+++ linux/arch/x86_64/kernel/genapic.c
@@ -32,30 +32,26 @@ extern struct genapic apic_cluster;
extern struct genapic apic_flat;
extern struct genapic apic_physflat;
-struct genapic *genapic = &apic_flat;
-
+struct genapic __read_mostly *genapic = &apic_flat;
/*
* Check the APIC IDs in bios_cpu_apicid and choose the APIC mode.
*/
void __init clustered_apic_check(void)
{
- long i;
- u8 clusters, max_cluster;
- u8 id;
- u8 cluster_cnt[NUM_APIC_CLUSTERS];
- int max_apic = 0;
+ u8 id, clusters, max_cluster, cluster_cnt[NUM_APIC_CLUSTERS];
+ int i, max_apic = 0;
-#if defined(CONFIG_ACPI)
+#ifdef CONFIG_ACPI
/*
* Some x86_64 machines use physical APIC mode regardless of how many
* procs/clusters are present (x86_64 ES7000 is an example).
*/
- if (acpi_fadt.revision > FADT2_REVISION_ID)
- if (acpi_fadt.force_apic_physical_destination_mode) {
- genapic = &apic_cluster;
- goto print;
- }
+ if (acpi_fadt.revision > FADT2_REVISION_ID &&
+ acpi_fadt.force_apic_physical_destination_mode) {
+ genapic = &apic_cluster;
+ goto print;
+ }
#endif
memset(cluster_cnt, 0, sizeof(cluster_cnt));
@@ -68,20 +64,17 @@ void __init clustered_apic_check(void)
cluster_cnt[APIC_CLUSTERID(id)]++;
}
- /* Don't use clustered mode on AMD platforms. */
+ /*
+ * Don't use clustered mode on AMD platforms, default
+ * to flat logical mode.
+ */
if (boot_cpu_data.x86_vendor == X86_VENDOR_AMD) {
- genapic = &apic_physflat;
-#ifndef CONFIG_HOTPLUG_CPU
- /* In the CPU hotplug case we cannot use broadcast mode
- because that opens a race when a CPU is removed.
- Stay at physflat mode in this case.
- It is bad to do this unconditionally though. Once
- we have ACPI platform support for CPU hotplug
- we should detect hotplug capablity from ACPI tables and
- only do this when really needed. -AK */
- if (max_apic <= 8)
- genapic = &apic_flat;
-#endif
+ /*
+ * Switch to physical flat mode if more than 8 APICs
+ * (In the case of 8 CPUs APIC ID goes from 0 to 7):
+ */
+ if (max_apic >= 8)
+ genapic = &apic_physflat;
goto print;
}
@@ -103,14 +96,9 @@ void __init clustered_apic_check(void)
* (We don't use lowest priority delivery + HW APIC IRQ steering, so
* can ignore the clustered logical case and go straight to physical.)
*/
- if (clusters <= 1 && max_cluster <= 8 && cluster_cnt[0] == max_cluster) {
-#ifdef CONFIG_HOTPLUG_CPU
- /* Don't use APIC shortcuts in CPU hotplug to avoid races */
- genapic = &apic_physflat;
-#else
+ if (clusters <= 1 && max_cluster <= 8 && cluster_cnt[0] == max_cluster)
genapic = &apic_flat;
-#endif
- } else
+ else
genapic = &apic_cluster;
print:
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists