e97ca8e5b8
GFP_THISNODE is for callers that implement their own clever fallback to remote nodes. It restricts the allocation to the specified node and does not invoke reclaim, assuming that the caller will take care of it when the fallback fails, e.g. through a subsequent allocation request without GFP_THISNODE set. However, many current GFP_THISNODE users only want the node exclusive aspect of the flag, without actually implementing their own fallback or triggering reclaim if necessary. This results in things like page migration failing prematurely even when there is easily reclaimable memory available, unless kswapd happens to be running already or a concurrent allocation attempt triggers the necessary reclaim. Convert all callsites that don't implement their own fallback strategy to __GFP_THISNODE. This restricts the allocation a single node too, but at the same time allows the allocator to enter the slowpath, wake kswapd, and invoke direct reclaim if necessary, to make the allocation happen when memory is full. Signed-off-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Rik van Riel <riel@redhat.com> Cc: Jan Stancek <jstancek@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
357 lines
7.8 KiB
C
357 lines
7.8 KiB
C
/*
|
|
* Copyright 2006-2008, IBM Corporation.
|
|
*
|
|
* This program is free software; you can redistribute it and/or
|
|
* modify it under the terms of the GNU General Public License
|
|
* as published by the Free Software Foundation; either version
|
|
* 2 of the License, or (at your option) any later version.
|
|
*/
|
|
|
|
#undef DEBUG
|
|
|
|
#include <linux/types.h>
|
|
#include <linux/kernel.h>
|
|
#include <linux/slab.h>
|
|
#include <linux/smp.h>
|
|
#include <linux/reboot.h>
|
|
#include <linux/kexec.h>
|
|
#include <linux/crash_dump.h>
|
|
|
|
#include <asm/kexec.h>
|
|
#include <asm/reg.h>
|
|
#include <asm/io.h>
|
|
#include <asm/prom.h>
|
|
#include <asm/machdep.h>
|
|
#include <asm/rtas.h>
|
|
#include <asm/cell-regs.h>
|
|
|
|
#include "ras.h"
|
|
|
|
|
|
static void dump_fir(int cpu)
|
|
{
|
|
struct cbe_pmd_regs __iomem *pregs = cbe_get_cpu_pmd_regs(cpu);
|
|
struct cbe_iic_regs __iomem *iregs = cbe_get_cpu_iic_regs(cpu);
|
|
|
|
if (pregs == NULL)
|
|
return;
|
|
|
|
/* Todo: do some nicer parsing of bits and based on them go down
|
|
* to other sub-units FIRs and not only IIC
|
|
*/
|
|
printk(KERN_ERR "Global Checkstop FIR : 0x%016llx\n",
|
|
in_be64(&pregs->checkstop_fir));
|
|
printk(KERN_ERR "Global Recoverable FIR : 0x%016llx\n",
|
|
in_be64(&pregs->checkstop_fir));
|
|
printk(KERN_ERR "Global MachineCheck FIR : 0x%016llx\n",
|
|
in_be64(&pregs->spec_att_mchk_fir));
|
|
|
|
if (iregs == NULL)
|
|
return;
|
|
printk(KERN_ERR "IOC FIR : 0x%016llx\n",
|
|
in_be64(&iregs->ioc_fir));
|
|
|
|
}
|
|
|
|
void cbe_system_error_exception(struct pt_regs *regs)
|
|
{
|
|
int cpu = smp_processor_id();
|
|
|
|
printk(KERN_ERR "System Error Interrupt on CPU %d !\n", cpu);
|
|
dump_fir(cpu);
|
|
dump_stack();
|
|
}
|
|
|
|
void cbe_maintenance_exception(struct pt_regs *regs)
|
|
{
|
|
int cpu = smp_processor_id();
|
|
|
|
/*
|
|
* Nothing implemented for the maintenance interrupt at this point
|
|
*/
|
|
|
|
printk(KERN_ERR "Unhandled Maintenance interrupt on CPU %d !\n", cpu);
|
|
dump_stack();
|
|
}
|
|
|
|
void cbe_thermal_exception(struct pt_regs *regs)
|
|
{
|
|
int cpu = smp_processor_id();
|
|
|
|
/*
|
|
* Nothing implemented for the thermal interrupt at this point
|
|
*/
|
|
|
|
printk(KERN_ERR "Unhandled Thermal interrupt on CPU %d !\n", cpu);
|
|
dump_stack();
|
|
}
|
|
|
|
static int cbe_machine_check_handler(struct pt_regs *regs)
|
|
{
|
|
int cpu = smp_processor_id();
|
|
|
|
printk(KERN_ERR "Machine Check Interrupt on CPU %d !\n", cpu);
|
|
dump_fir(cpu);
|
|
|
|
/* No recovery from this code now, lets continue */
|
|
return 0;
|
|
}
|
|
|
|
struct ptcal_area {
|
|
struct list_head list;
|
|
int nid;
|
|
int order;
|
|
struct page *pages;
|
|
};
|
|
|
|
static LIST_HEAD(ptcal_list);
|
|
|
|
static int ptcal_start_tok, ptcal_stop_tok;
|
|
|
|
static int __init cbe_ptcal_enable_on_node(int nid, int order)
|
|
{
|
|
struct ptcal_area *area;
|
|
int ret = -ENOMEM;
|
|
unsigned long addr;
|
|
|
|
if (is_kdump_kernel())
|
|
rtas_call(ptcal_stop_tok, 1, 1, NULL, nid);
|
|
|
|
area = kmalloc(sizeof(*area), GFP_KERNEL);
|
|
if (!area)
|
|
goto out_err;
|
|
|
|
area->nid = nid;
|
|
area->order = order;
|
|
area->pages = alloc_pages_exact_node(area->nid,
|
|
GFP_KERNEL|__GFP_THISNODE,
|
|
area->order);
|
|
|
|
if (!area->pages) {
|
|
printk(KERN_WARNING "%s: no page on node %d\n",
|
|
__func__, area->nid);
|
|
goto out_free_area;
|
|
}
|
|
|
|
/*
|
|
* We move the ptcal area to the middle of the allocated
|
|
* page, in order to avoid prefetches in memcpy and similar
|
|
* functions stepping on it.
|
|
*/
|
|
addr = __pa(page_address(area->pages)) + (PAGE_SIZE >> 1);
|
|
printk(KERN_DEBUG "%s: enabling PTCAL on node %d address=0x%016lx\n",
|
|
__func__, area->nid, addr);
|
|
|
|
ret = -EIO;
|
|
if (rtas_call(ptcal_start_tok, 3, 1, NULL, area->nid,
|
|
(unsigned int)(addr >> 32),
|
|
(unsigned int)(addr & 0xffffffff))) {
|
|
printk(KERN_ERR "%s: error enabling PTCAL on node %d!\n",
|
|
__func__, nid);
|
|
goto out_free_pages;
|
|
}
|
|
|
|
list_add(&area->list, &ptcal_list);
|
|
|
|
return 0;
|
|
|
|
out_free_pages:
|
|
__free_pages(area->pages, area->order);
|
|
out_free_area:
|
|
kfree(area);
|
|
out_err:
|
|
return ret;
|
|
}
|
|
|
|
static int __init cbe_ptcal_enable(void)
|
|
{
|
|
const u32 *size;
|
|
struct device_node *np;
|
|
int order, found_mic = 0;
|
|
|
|
np = of_find_node_by_path("/rtas");
|
|
if (!np)
|
|
return -ENODEV;
|
|
|
|
size = of_get_property(np, "ibm,cbe-ptcal-size", NULL);
|
|
if (!size) {
|
|
of_node_put(np);
|
|
return -ENODEV;
|
|
}
|
|
|
|
pr_debug("%s: enabling PTCAL, size = 0x%x\n", __func__, *size);
|
|
order = get_order(*size);
|
|
of_node_put(np);
|
|
|
|
/* support for malta device trees, with be@/mic@ nodes */
|
|
for_each_node_by_type(np, "mic-tm") {
|
|
cbe_ptcal_enable_on_node(of_node_to_nid(np), order);
|
|
found_mic = 1;
|
|
}
|
|
|
|
if (found_mic)
|
|
return 0;
|
|
|
|
/* support for older device tree - use cpu nodes */
|
|
for_each_node_by_type(np, "cpu") {
|
|
const u32 *nid = of_get_property(np, "node-id", NULL);
|
|
if (!nid) {
|
|
printk(KERN_ERR "%s: node %s is missing node-id?\n",
|
|
__func__, np->full_name);
|
|
continue;
|
|
}
|
|
cbe_ptcal_enable_on_node(*nid, order);
|
|
found_mic = 1;
|
|
}
|
|
|
|
return found_mic ? 0 : -ENODEV;
|
|
}
|
|
|
|
static int cbe_ptcal_disable(void)
|
|
{
|
|
struct ptcal_area *area, *tmp;
|
|
int ret = 0;
|
|
|
|
pr_debug("%s: disabling PTCAL\n", __func__);
|
|
|
|
list_for_each_entry_safe(area, tmp, &ptcal_list, list) {
|
|
/* disable ptcal on this node */
|
|
if (rtas_call(ptcal_stop_tok, 1, 1, NULL, area->nid)) {
|
|
printk(KERN_ERR "%s: error disabling PTCAL "
|
|
"on node %d!\n", __func__,
|
|
area->nid);
|
|
ret = -EIO;
|
|
continue;
|
|
}
|
|
|
|
/* ensure we can access the PTCAL area */
|
|
memset(page_address(area->pages), 0,
|
|
1 << (area->order + PAGE_SHIFT));
|
|
|
|
/* clean up */
|
|
list_del(&area->list);
|
|
__free_pages(area->pages, area->order);
|
|
kfree(area);
|
|
}
|
|
|
|
return ret;
|
|
}
|
|
|
|
static int cbe_ptcal_notify_reboot(struct notifier_block *nb,
|
|
unsigned long code, void *data)
|
|
{
|
|
return cbe_ptcal_disable();
|
|
}
|
|
|
|
static void cbe_ptcal_crash_shutdown(void)
|
|
{
|
|
cbe_ptcal_disable();
|
|
}
|
|
|
|
static struct notifier_block cbe_ptcal_reboot_notifier = {
|
|
.notifier_call = cbe_ptcal_notify_reboot
|
|
};
|
|
|
|
#ifdef CONFIG_PPC_IBM_CELL_RESETBUTTON
|
|
static int sysreset_hack;
|
|
|
|
static int __init cbe_sysreset_init(void)
|
|
{
|
|
struct cbe_pmd_regs __iomem *regs;
|
|
|
|
sysreset_hack = of_machine_is_compatible("IBM,CBPLUS-1.0");
|
|
if (!sysreset_hack)
|
|
return 0;
|
|
|
|
regs = cbe_get_cpu_pmd_regs(0);
|
|
if (!regs)
|
|
return 0;
|
|
|
|
/* Enable JTAG system-reset hack */
|
|
out_be32(®s->fir_mode_reg,
|
|
in_be32(®s->fir_mode_reg) |
|
|
CBE_PMD_FIR_MODE_M8);
|
|
|
|
return 0;
|
|
}
|
|
device_initcall(cbe_sysreset_init);
|
|
|
|
int cbe_sysreset_hack(void)
|
|
{
|
|
struct cbe_pmd_regs __iomem *regs;
|
|
|
|
/*
|
|
* The BMC can inject user triggered system reset exceptions,
|
|
* but cannot set the system reset reason in srr1,
|
|
* so check an extra register here.
|
|
*/
|
|
if (sysreset_hack && (smp_processor_id() == 0)) {
|
|
regs = cbe_get_cpu_pmd_regs(0);
|
|
if (!regs)
|
|
return 0;
|
|
if (in_be64(®s->ras_esc_0) & 0x0000ffff) {
|
|
out_be64(®s->ras_esc_0, 0);
|
|
return 0;
|
|
}
|
|
}
|
|
return 1;
|
|
}
|
|
#endif /* CONFIG_PPC_IBM_CELL_RESETBUTTON */
|
|
|
|
int __init cbe_ptcal_init(void)
|
|
{
|
|
int ret;
|
|
ptcal_start_tok = rtas_token("ibm,cbe-start-ptcal");
|
|
ptcal_stop_tok = rtas_token("ibm,cbe-stop-ptcal");
|
|
|
|
if (ptcal_start_tok == RTAS_UNKNOWN_SERVICE
|
|
|| ptcal_stop_tok == RTAS_UNKNOWN_SERVICE)
|
|
return -ENODEV;
|
|
|
|
ret = register_reboot_notifier(&cbe_ptcal_reboot_notifier);
|
|
if (ret)
|
|
goto out1;
|
|
|
|
ret = crash_shutdown_register(&cbe_ptcal_crash_shutdown);
|
|
if (ret)
|
|
goto out2;
|
|
|
|
return cbe_ptcal_enable();
|
|
|
|
out2:
|
|
unregister_reboot_notifier(&cbe_ptcal_reboot_notifier);
|
|
out1:
|
|
printk(KERN_ERR "Can't disable PTCAL, so not enabling\n");
|
|
return ret;
|
|
}
|
|
|
|
arch_initcall(cbe_ptcal_init);
|
|
|
|
void __init cbe_ras_init(void)
|
|
{
|
|
unsigned long hid0;
|
|
|
|
/*
|
|
* Enable System Error & thermal interrupts and wakeup conditions
|
|
*/
|
|
|
|
hid0 = mfspr(SPRN_HID0);
|
|
hid0 |= HID0_CBE_THERM_INT_EN | HID0_CBE_THERM_WAKEUP |
|
|
HID0_CBE_SYSERR_INT_EN | HID0_CBE_SYSERR_WAKEUP;
|
|
mtspr(SPRN_HID0, hid0);
|
|
mb();
|
|
|
|
/*
|
|
* Install machine check handler. Leave setting of precise mode to
|
|
* what the firmware did for now
|
|
*/
|
|
ppc_md.machine_check_exception = cbe_machine_check_handler;
|
|
mb();
|
|
|
|
/*
|
|
* For now, we assume that IOC_FIR is already set to forward some
|
|
* error conditions to the System Error handler. If that is not true
|
|
* then it will have to be fixed up here.
|
|
*/
|
|
}
|