From d596043d71ff0d7b3d0bead19b1d68c55f003093 Mon Sep 17 00:00:00 2001 From: "Darrick J. Wong" Date: Wed, 30 Jun 2010 17:45:19 -0700 Subject: x86, Calgary: Limit the max PHB number to 256 The x3950 family can have as many as 256 PCI buses in a single system, so change the limits to the maximum. Since there can only be 256 PCI buses in one domain, we no longer need the BUG_ON check. Signed-off-by: Darrick J. Wong LKML-Reference: <20100701004519.GQ15515@tux1.beaverton.ibm.com> Signed-off-by: H. Peter Anvin --- arch/x86/kernel/pci-calgary_64.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/pci-calgary_64.c b/arch/x86/kernel/pci-calgary_64.c index 0b96b5589f08..078d4ec1a9d9 100644 --- a/arch/x86/kernel/pci-calgary_64.c +++ b/arch/x86/kernel/pci-calgary_64.c @@ -110,7 +110,7 @@ int use_calgary __read_mostly = 0; * x3950 (PCIE): 8 chassis, 32 PHBs per chassis = 256 * x3950 (PCIX): 8 chassis, 16 PHBs per chassis = 128 */ -#define MAX_PHB_BUS_NUM 384 +#define MAX_PHB_BUS_NUM 256 #define PHBS_PER_CALGARY 4 @@ -1056,8 +1056,6 @@ static int __init calgary_init_one(struct pci_dev *dev) struct iommu_table *tbl; int ret; - BUG_ON(dev->bus->number >= MAX_PHB_BUS_NUM); - bbar = busno_to_bbar(dev->bus->number); ret = calgary_setup_tar(dev, bbar); if (ret) -- cgit v1.2.2 From b945d6b2554d550fe95caadc61e521c0ad71fb9c Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Sat, 29 May 2010 15:31:43 +0200 Subject: rbtree: Undo augmented trees performance damage and regression Reimplement augmented RB-trees without sprinkling extra branches all over the RB-tree code (which lives in the scheduler hot path). This approach is 'borrowed' from Fabio's BFQ implementation and relies on traversing the rebalance path after the RB-tree-op to correct the heap property for insertion/removal and make up for the damage done by the tree rotations. For insertion the rebalance path is trivially that from the new node upwards to the root, for removal it is that from the deepest node in the path from the to be removed node that will still be around after the removal. [ This patch also fixes a video driver regression reported by Ali Gholami Rudi - the memtype->subtree_max_end was updated incorrectly. ] Acked-by: Suresh Siddha Acked-by: Venkatesh Pallipadi Signed-off-by: Peter Zijlstra Tested-by: Ali Gholami Rudi Cc: Fabio Checconi Cc: "H. Peter Anvin" Cc: Andrew Morton Cc: Linus Torvalds LKML-Reference: <1275414172.27810.27961.camel@twins> Signed-off-by: Ingo Molnar --- arch/x86/mm/pat_rbtree.c | 34 ++++++---------------------------- 1 file changed, 6 insertions(+), 28 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/mm/pat_rbtree.c b/arch/x86/mm/pat_rbtree.c index f20eeec85a86..8acaddd0fb21 100644 --- a/arch/x86/mm/pat_rbtree.c +++ b/arch/x86/mm/pat_rbtree.c @@ -34,8 +34,7 @@ * memtype_lock protects the rbtree. */ -static void memtype_rb_augment_cb(struct rb_node *node); -static struct rb_root memtype_rbroot = RB_AUGMENT_ROOT(&memtype_rb_augment_cb); +static struct rb_root memtype_rbroot = RB_ROOT; static int is_node_overlap(struct memtype *node, u64 start, u64 end) { @@ -56,7 +55,7 @@ static u64 get_subtree_max_end(struct rb_node *node) } /* Update 'subtree_max_end' for a node, based on node and its children */ -static void update_node_max_end(struct rb_node *node) +static void memtype_rb_augment_cb(struct rb_node *node, void *__unused) { struct memtype *data; u64 max_end, child_max_end; @@ -78,25 +77,6 @@ static void update_node_max_end(struct rb_node *node) data->subtree_max_end = max_end; } -/* Update 'subtree_max_end' for a node and all its ancestors */ -static void update_path_max_end(struct rb_node *node) -{ - u64 old_max_end, new_max_end; - - while (node) { - struct memtype *data = container_of(node, struct memtype, rb); - - old_max_end = data->subtree_max_end; - update_node_max_end(node); - new_max_end = data->subtree_max_end; - - if (new_max_end == old_max_end) - break; - - node = rb_parent(node); - } -} - /* Find the first (lowest start addr) overlapping range from rb tree */ static struct memtype *memtype_rb_lowest_match(struct rb_root *root, u64 start, u64 end) @@ -190,12 +170,6 @@ failure: return -EBUSY; } -static void memtype_rb_augment_cb(struct rb_node *node) -{ - if (node) - update_path_max_end(node); -} - static void memtype_rb_insert(struct rb_root *root, struct memtype *newdata) { struct rb_node **node = &(root->rb_node); @@ -213,6 +187,7 @@ static void memtype_rb_insert(struct rb_root *root, struct memtype *newdata) rb_link_node(&newdata->rb, parent, node); rb_insert_color(&newdata->rb, root); + rb_augment_insert(&newdata->rb, memtype_rb_augment_cb, NULL); } int rbt_memtype_check_insert(struct memtype *new, unsigned long *ret_type) @@ -234,13 +209,16 @@ int rbt_memtype_check_insert(struct memtype *new, unsigned long *ret_type) struct memtype *rbt_memtype_erase(u64 start, u64 end) { + struct rb_node *deepest; struct memtype *data; data = memtype_rb_exact_match(&memtype_rbroot, start, end); if (!data) goto out; + deepest = rb_augment_erase_begin(&data->rb); rb_erase(&data->rb, &memtype_rbroot); + rb_augment_erase_end(deepest, memtype_rb_augment_cb, NULL); out: return data; } -- cgit v1.2.2 From da38f43859467a8048365b9e1cce99ccbc62b6e2 Mon Sep 17 00:00:00 2001 From: Avi Kivity Date: Tue, 6 Jul 2010 11:30:49 +0300 Subject: KVM: VMX: Fix host MSR_KERNEL_GS_BASE corruption enter_lmode() and exit_lmode() modify the guest's EFER.LMA before calling vmx_set_efer(). However, the latter function depends on the value of EFER.LMA to determine whether MSR_KERNEL_GS_BASE needs reloading, via vmx_load_host_state(). With EFER.LMA changing under its feet, it took the wrong choice and corrupted userspace's %gs. This causes 32-on-64 host userspace to fault. Fix not touching EFER.LMA; instead ask vmx_set_efer() to change it. Signed-off-by: Avi Kivity --- arch/x86/kvm/vmx.c | 7 ++----- 1 file changed, 2 insertions(+), 5 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c index 859a01a07dbf..ee03679efe78 100644 --- a/arch/x86/kvm/vmx.c +++ b/arch/x86/kvm/vmx.c @@ -1744,18 +1744,15 @@ static void enter_lmode(struct kvm_vcpu *vcpu) (guest_tr_ar & ~AR_TYPE_MASK) | AR_TYPE_BUSY_64_TSS); } - vcpu->arch.efer |= EFER_LMA; - vmx_set_efer(vcpu, vcpu->arch.efer); + vmx_set_efer(vcpu, vcpu->arch.efer | EFER_LMA); } static void exit_lmode(struct kvm_vcpu *vcpu) { - vcpu->arch.efer &= ~EFER_LMA; - vmcs_write32(VM_ENTRY_CONTROLS, vmcs_read32(VM_ENTRY_CONTROLS) & ~VM_ENTRY_IA32E_MODE); - vmx_set_efer(vcpu, vcpu->arch.efer); + vmx_set_efer(vcpu, vcpu->arch.efer & ~EFER_LMA); } #endif -- cgit v1.2.2 From 72550b3ae545c75897c769d43d62d4be3f3d48fe Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Wed, 7 Jul 2010 16:57:46 -0700 Subject: x86, platform: Export x86_platform to modules Export x86_platform to modules in preparation of using it for i8042 discovery control. Signed-off-by: H. Peter Anvin LKML-Reference: <1278342202-10973-1-git-send-email-feng.tang@intel.com> Cc: Thomas Gleixner Cc: Feng Tang Cc: Dmitry Torokhov --- arch/x86/kernel/x86_init.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'arch/x86') diff --git a/arch/x86/kernel/x86_init.c b/arch/x86/kernel/x86_init.c index 61a1e8c7e19f..ebfb8e4c9f22 100644 --- a/arch/x86/kernel/x86_init.c +++ b/arch/x86/kernel/x86_init.c @@ -5,6 +5,7 @@ */ #include #include +#include #include #include @@ -94,3 +95,5 @@ struct x86_platform_ops x86_platform = { .is_untracked_pat_range = is_ISA_range, .nmi_init = default_nmi_init }; + +EXPORT_SYMBOL_GPL(x86_platform); -- cgit v1.2.2 From c516ac583973196162b1ba7e4d597d6f6892dac0 Mon Sep 17 00:00:00 2001 From: Feng Tang Date: Mon, 5 Jul 2010 23:03:18 +0800 Subject: x86: Add i8042 pre-detection hook to x86_platform_ops Some x86 platforms like Intel MID platforms don't have i8042 controllers, and i8042 driver's probe to some legacy IO ports may hang the MID processor. With this hook, i8042 driver can runtime check and skip the probe when the pretection fail which also saves some probe time [ hpa note: this is currently a compile-time check, which breaks the i386 allyesconfig build. This patch series thus does fix a regression. ] Signed-off-by: Feng Tang LKML-Reference: <1278342202-10973-2-git-send-email-feng.tang@intel.com> Acked-by: Dmitry Torokhov Signed-off-by: H. Peter Anvin --- arch/x86/include/asm/x86_init.h | 2 ++ arch/x86/kernel/x86_init.c | 4 +++- 2 files changed, 5 insertions(+), 1 deletion(-) (limited to 'arch/x86') diff --git a/arch/x86/include/asm/x86_init.h b/arch/x86/include/asm/x86_init.h index 519b54327d75..baa579c8e038 100644 --- a/arch/x86/include/asm/x86_init.h +++ b/arch/x86/include/asm/x86_init.h @@ -142,6 +142,7 @@ struct x86_cpuinit_ops { * @set_wallclock: set time back to HW clock * @is_untracked_pat_range exclude from PAT logic * @nmi_init enable NMI on cpus + * @i8042_detect pre-detect if i8042 controller exists */ struct x86_platform_ops { unsigned long (*calibrate_tsc)(void); @@ -150,6 +151,7 @@ struct x86_platform_ops { void (*iommu_shutdown)(void); bool (*is_untracked_pat_range)(u64 start, u64 end); void (*nmi_init)(void); + int (*i8042_detect)(void); }; extern struct x86_init_ops x86_init; diff --git a/arch/x86/kernel/x86_init.c b/arch/x86/kernel/x86_init.c index ebfb8e4c9f22..cd6da6bf3eca 100644 --- a/arch/x86/kernel/x86_init.c +++ b/arch/x86/kernel/x86_init.c @@ -86,6 +86,7 @@ struct x86_cpuinit_ops x86_cpuinit __cpuinitdata = { }; static void default_nmi_init(void) { }; +static int default_i8042_detect(void) { return 1; }; struct x86_platform_ops x86_platform = { .calibrate_tsc = native_calibrate_tsc, @@ -93,7 +94,8 @@ struct x86_platform_ops x86_platform = { .set_wallclock = mach_set_rtc_mmss, .iommu_shutdown = iommu_shutdown_noop, .is_untracked_pat_range = is_ISA_range, - .nmi_init = default_nmi_init + .nmi_init = default_nmi_init, + .i8042_detect = default_i8042_detect }; EXPORT_SYMBOL_GPL(x86_platform); -- cgit v1.2.2 From 6d2cce62017efe957e34cfcbba23861b7671980b Mon Sep 17 00:00:00 2001 From: Feng Tang Date: Mon, 5 Jul 2010 23:03:19 +0800 Subject: x86, mrst: Add i8042_detect API for Moorestwon platform It will just return 0 as there is no i8042 controller Signed-off-by: Feng Tang LKML-Reference: <1278342202-10973-3-git-send-email-feng.tang@intel.com> Acked-by: Dmitry Torokhov Signed-off-by: H. Peter Anvin --- arch/x86/kernel/mrst.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'arch/x86') diff --git a/arch/x86/kernel/mrst.c b/arch/x86/kernel/mrst.c index e796448f0eb5..5915e0b33303 100644 --- a/arch/x86/kernel/mrst.c +++ b/arch/x86/kernel/mrst.c @@ -216,6 +216,12 @@ static void __init mrst_setup_boot_clock(void) setup_boot_APIC_clock(); }; +/* MID systems don't have i8042 controller */ +static int mrst_i8042_detect(void) +{ + return 0; +} + /* * Moorestown specific x86_init function overrides and early setup * calls. @@ -233,6 +239,7 @@ void __init x86_mrst_early_setup(void) x86_cpuinit.setup_percpu_clockev = mrst_setup_secondary_clock; x86_platform.calibrate_tsc = mrst_calibrate_tsc; + x86_platform.i8042_detect = mrst_i8042_detect; x86_init.pci.init = pci_mrst_init; x86_init.pci.fixup_irqs = x86_init_noop; -- cgit v1.2.2 From 91546356d0e550fa23abf7f4b04a903c2855761f Mon Sep 17 00:00:00 2001 From: Xiao Guangrong Date: Wed, 30 Jun 2010 16:04:06 +0800 Subject: KVM: MMU: flush remote tlbs when overwriting spte with different pfn After remove a rmap, we should flush all vcpu's tlb Signed-off-by: Xiao Guangrong Signed-off-by: Marcelo Tosatti --- arch/x86/kvm/mmu.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch/x86') diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c index a6f695d76928..3699613e8830 100644 --- a/arch/x86/kvm/mmu.c +++ b/arch/x86/kvm/mmu.c @@ -1879,6 +1879,8 @@ static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *sptep, pgprintk("hfn old %lx new %lx\n", spte_to_pfn(*sptep), pfn); rmap_remove(vcpu->kvm, sptep); + __set_spte(sptep, shadow_trap_nonpresent_pte); + kvm_flush_remote_tlbs(vcpu->kvm); } else was_rmapped = 1; } -- cgit v1.2.2 From 08be97962bf338161325d4901642f956ce8c1adb Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 14 Jul 2010 21:36:27 +0200 Subject: x86: Force HPET readback_cmp for all ATI chipsets commit 30a564be (x86, hpet: Restrict read back to affected ATI chipset) restricted the workaround for the HPET bug to SMX00 chipsets. This was reasonable as those were the only ones against which we ever got a bug report. Stephan Wolf reported now that this patch breaks his IXP400 based machine. Though it's confirmed to work on other IXP400 based systems. To error out on the safe side, we force the HPET readback workaround for all ATI SMbus class chipsets. Reported-by: Stephan Wolf LKML-Reference: Signed-off-by: Thomas Gleixner Tested-by: Stephan Wolf Acked-by: Borislav Petkov --- arch/x86/kernel/early-quirks.c | 18 ++++++++++++++++++ arch/x86/kernel/quirks.c | 5 ----- 2 files changed, 18 insertions(+), 5 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/early-quirks.c b/arch/x86/kernel/early-quirks.c index ebdb85cf2686..e5cc7e82e60d 100644 --- a/arch/x86/kernel/early-quirks.c +++ b/arch/x86/kernel/early-quirks.c @@ -18,6 +18,7 @@ #include #include #include +#include static void __init fix_hypertransport_config(int num, int slot, int func) { @@ -191,6 +192,21 @@ static void __init ati_bugs_contd(int num, int slot, int func) } #endif +/* + * Force the read back of the CMP register in hpet_next_event() + * to work around the problem that the CMP register write seems to be + * delayed. See hpet_next_event() for details. + * + * We do this on all SMBUS incarnations for now until we have more + * information about the affected chipsets. + */ +static void __init ati_hpet_bugs(int num, int slot, int func) +{ +#ifdef CONFIG_HPET_TIMER + hpet_readback_cmp = 1; +#endif +} + #define QFLAG_APPLY_ONCE 0x1 #define QFLAG_APPLIED 0x2 #define QFLAG_DONE (QFLAG_APPLY_ONCE|QFLAG_APPLIED) @@ -220,6 +236,8 @@ static struct chipset early_qrk[] __initdata = { PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs }, { PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_SBX00_SMBUS, PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs_contd }, + { PCI_VENDOR_ID_ATI, PCI_ANY_ID, + PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_hpet_bugs }, {} }; diff --git a/arch/x86/kernel/quirks.c b/arch/x86/kernel/quirks.c index e72d3fc6547d..939b9e98245f 100644 --- a/arch/x86/kernel/quirks.c +++ b/arch/x86/kernel/quirks.c @@ -498,15 +498,10 @@ void force_hpet_resume(void) * See erratum #27 (Misinterpreted MSI Requests May Result in * Corrupted LPC DMA Data) in AMD Publication #46837, * "SB700 Family Product Errata", Rev. 1.0, March 2010. - * - * Also force the read back of the CMP register in hpet_next_event() - * to work around the problem that the CMP register write seems to be - * delayed. See hpet_next_event() for details. */ static void force_disable_hpet_msi(struct pci_dev *unused) { hpet_msi_disable = 1; - hpet_readback_cmp = 1; } DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_SBX00_SMBUS, -- cgit v1.2.2 From 58c84eda07560a6b75b03e8d3b26d6eddfc14011 Mon Sep 17 00:00:00 2001 From: Bjorn Helgaas Date: Thu, 15 Jul 2010 09:41:42 -0600 Subject: PCI: fall back to original BIOS BAR addresses If we fail to assign resources to a PCI BAR, this patch makes us try the original address from BIOS rather than leaving it disabled. Linux tries to make sure all PCI device BARs are inside the upstream PCI host bridge or P2P bridge apertures, reassigning BARs if necessary. Windows does similar reassignment. Before this patch, if we could not move a BAR into an aperture, we left the resource unassigned, i.e., at address zero. Windows leaves such BARs at the original BIOS addresses, and this patch makes Linux do the same. This is a bit ugly because we disable the resource long before we try to reassign it, so we have to keep track of the BIOS BAR address somewhere. For lack of a better place, I put it in the struct pci_dev. I think it would be cleaner to attempt the assignment immediately when the claim fails, so we could easily remember the original address. But we currently claim motherboard resources in the middle, after attempting to claim PCI resources and before assigning new PCI resources, and changing that is a fairly big job. Addresses https://bugzilla.kernel.org/show_bug.cgi?id=16263 Reported-by: Andrew Tested-by: Andrew Signed-off-by: Bjorn Helgaas Signed-off-by: Jesse Barnes --- arch/x86/pci/i386.c | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/x86') diff --git a/arch/x86/pci/i386.c b/arch/x86/pci/i386.c index 6fdb3ec30c31..55253095be84 100644 --- a/arch/x86/pci/i386.c +++ b/arch/x86/pci/i386.c @@ -184,6 +184,7 @@ static void __init pcibios_allocate_resources(int pass) idx, r, disabled, pass); if (pci_claim_resource(dev, idx) < 0) { /* We'll assign a new address later */ + dev->fw_addr[idx] = r->start; r->end -= r->start; r->start = 0; } -- cgit v1.2.2 From fd19dce7ac07973f700b0f13fb7f94b951414a4c Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Thu, 15 Jul 2010 00:00:59 -0700 Subject: x86: Fix x2apic preenabled system with kexec Found one x2apic system kexec loop test failed when CONFIG_NMI_WATCHDOG=y (old) or CONFIG_LOCKUP_DETECTOR=y (current tip) first kernel can kexec second kernel, but second kernel can not kexec third one. it can be duplicated on another system with BIOS preenabled x2apic. First kernel can not kexec second kernel. It turns out, when kernel boot with pre-enabled x2apic, it will not execute disable_local_APIC on shutdown path. when init_apic_mappings() is called in setup_arch, it will skip setting of apic_phys when x2apic_mode is set. ( x2apic_mode is much early check_x2apic()) Then later, disable_local_APIC() will bail out early because !apic_phys. So check !x2apic_mode in x2apic_mode in disable_local_APIC with !apic_phys. another solution could be updating init_apic_mappings() to set apic_phys even for preenabled x2apic system. Actually even for x2apic system, that lapic address is mapped already in early stage. BTW: is there any x2apic preenabled system with apicid of boot cpu > 255? Signed-off-by: Yinghai Lu LKML-Reference: <4C3EB22B.3000701@kernel.org> Acked-by: Suresh Siddha Cc: stable@kernel.org Signed-off-by: H. Peter Anvin --- arch/x86/kernel/apic/apic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index c02cc692985c..a96489ee6cab 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -921,7 +921,7 @@ void disable_local_APIC(void) unsigned int value; /* APIC hasn't been mapped yet */ - if (!apic_phys) + if (!x2apic_mode && !apic_phys) return; clear_local_APIC(); -- cgit v1.2.2 From f82c3d71d6fd2e6a3e3416f09099e29087e39abf Mon Sep 17 00:00:00 2001 From: Jacob Pan Date: Fri, 16 Jul 2010 11:58:26 -0700 Subject: x86, pci, mrst: Add extra sanity check in walking the PCI extended cap chain The fixed bar capability structure is searched in PCI extended configuration space. We need to make sure there is a valid capability ID to begin with otherwise, the search code may stuck in a infinite loop which results in boot hang. This patch adds additional check for cap ID 0, which is also invalid, and indicates end of chain. End of chain is supposed to have all fields zero, but that doesn't seem to always be the case in the field. Suggested-by: "H. Peter Anvin" Signed-off-by: Jacob Pan Reviewed-by: Jesse Barnes LKML-Reference: <1279306706-27087-1-git-send-email-jacob.jun.pan@linux.intel.com> Signed-off-by: H. Peter Anvin --- arch/x86/pci/mrst.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/pci/mrst.c b/arch/x86/pci/mrst.c index 7ef3a2735df3..cb29191cee58 100644 --- a/arch/x86/pci/mrst.c +++ b/arch/x86/pci/mrst.c @@ -66,8 +66,9 @@ static int fixed_bar_cap(struct pci_bus *bus, unsigned int devfn) devfn, pos, 4, &pcie_cap)) return 0; - if (pcie_cap == 0xffffffff) - return 0; + if (PCI_EXT_CAP_ID(pcie_cap) == 0x0000 || + PCI_EXT_CAP_ID(pcie_cap) == 0xffff) + break; if (PCI_EXT_CAP_ID(pcie_cap) == PCI_EXT_CAP_ID_VNDR) { raw_pci_ext_ops->read(pci_domain_nr(bus), bus->number, @@ -76,7 +77,7 @@ static int fixed_bar_cap(struct pci_bus *bus, unsigned int devfn) return pos; } - pos = pcie_cap >> 20; + pos = PCI_EXT_CAP_NEXT(pcie_cap); } return 0; -- cgit v1.2.2 From a197479848a2f1a2a5c07cffa6c31ab5e8c82797 Mon Sep 17 00:00:00 2001 From: Roland McGrath Date: Fri, 16 Jul 2010 18:17:12 -0700 Subject: x86: kprobes: fix swapped segment registers in kretprobe In commit f007ea26, the order of the %es and %ds segment registers got accidentally swapped, so synthesized 'struct pt_regs' frames have the two values inverted. It's almost sure that these values never matter, and that they also never differ. But wrong is wrong. Signed-off-by: Roland McGrath --- arch/x86/kernel/kprobes.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/kprobes.c b/arch/x86/kernel/kprobes.c index 345a4b1fe144..675879b65ce6 100644 --- a/arch/x86/kernel/kprobes.c +++ b/arch/x86/kernel/kprobes.c @@ -640,8 +640,8 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) /* Skip cs, ip, orig_ax and gs. */ \ " subl $16, %esp\n" \ " pushl %fs\n" \ - " pushl %ds\n" \ " pushl %es\n" \ + " pushl %ds\n" \ " pushl %eax\n" \ " pushl %ebp\n" \ " pushl %edi\n" \ -- cgit v1.2.2 From 7f8275d0d660c146de6ee3017e1e2e594c49e820 Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Mon, 19 Jul 2010 14:56:17 +1000 Subject: mm: add context argument to shrinker callback The current shrinker implementation requires the registered callback to have global state to work from. This makes it difficult to shrink caches that are not global (e.g. per-filesystem caches). Pass the shrinker structure to the callback so that users can embed the shrinker structure in the context the shrinker needs to operate on and get back to it in the callback via container_of(). Signed-off-by: Dave Chinner Reviewed-by: Christoph Hellwig --- arch/x86/kvm/mmu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86') diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c index 3699613e8830..b1ed0a1a5913 100644 --- a/arch/x86/kvm/mmu.c +++ b/arch/x86/kvm/mmu.c @@ -2926,7 +2926,7 @@ static int kvm_mmu_remove_some_alloc_mmu_pages(struct kvm *kvm) return kvm_mmu_zap_page(kvm, page) + 1; } -static int mmu_shrink(int nr_to_scan, gfp_t gfp_mask) +static int mmu_shrink(struct shrinker *shrink, int nr_to_scan, gfp_t gfp_mask) { struct kvm *kvm; struct kvm *kvm_freed = NULL; -- cgit v1.2.2 From 9aebbdb637a73a6092e1456ebb4a2df32cc1f611 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 20 Jul 2010 13:24:30 -0700 Subject: x86, numa: fix boot without RAM on node0 again Commit e534c7c5f8d6 ("numa: x86_64: use generic percpu var numa_node_id() implementation") broke numa systems that don't have ram on node0 when MEMORY_HOTPLUG is enabled, because cpu_up() will call cpu_to_node() before per_cpu(numa_node) is setup for APs. When Node0 doesn't have RAM, on x86, cpus already round it to nearest node with RAM in x86_cpu_to_node_map. and per_cpu(numa_node) is not set up until in c_init for APs. When later cpu_up() calling cpu_to_node() will get 0 again, and make it online even there is no RAM on node0. so later all APs can not booted up, and later will have panic. [ 1.611101] On node 0 totalpages: 0 ......... [ 2.608558] On node 0 totalpages: 0 [ 2.612065] Brought up 1 CPUs [ 2.615199] Total of 1 processors activated (3990.31 BogoMIPS). ... 93.225341] calling loop_init+0x0/0x1a4 @ 1 [ 93.229314] PERCPU: allocation failed, size=80 align=8, failed to populate [ 93.246539] Pid: 1, comm: swapper Tainted: G W 2.6.35-rc4-tip-yh-04371-gd64e6c4-dirty #354 [ 93.264621] Call Trace: [ 93.266533] [] pcpu_alloc+0x83a/0x8e7 [ 93.270710] [] __alloc_percpu+0x10/0x12 [ 93.285849] [] alloc_disk_node+0x94/0x16d [ 93.291811] [] alloc_disk+0x11/0x13 [ 93.306157] [] loop_alloc+0xa7/0x180 [ 93.310538] [] loop_init+0x9b/0x1a4 [ 93.324909] [] ? loop_init+0x0/0x1a4 [ 93.329650] [] do_one_initcall+0x57/0x136 [ 93.345197] [] kernel_init+0x184/0x20e [ 93.348146] [] kernel_thread_helper+0x4/0x10 [ 93.365194] [] ? restore_args+0x0/0x30 [ 93.369305] [] ? kernel_init+0x0/0x20e [ 93.386011] [] ? kernel_thread_helper+0x0/0x10 [ 93.392047] loop: out of memory ... Try to assign per_cpu(numa_node) early [akpm@linux-foundation.org: tidy up code comment] Signed-off-by: Yinghai Cc: Thomas Gleixner Cc: Ingo Molnar Cc: "H. Peter Anvin" Cc: Tejun Heo Cc: Denys Vlasenko Acked-by: Lee Schermerhorn Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/x86/kernel/setup_percpu.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index de3b63ae3da2..690c2c09faf3 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -238,6 +238,15 @@ void __init setup_per_cpu_areas(void) #ifdef CONFIG_NUMA per_cpu(x86_cpu_to_node_map, cpu) = early_per_cpu_map(x86_cpu_to_node_map, cpu); + /* + * Ensure taht the boot cpu numa_node is correct when the boot + * cpu is on a node that doesn't have memory installed. + * Also cpu_up() will call cpu_to_node() for APs when + * MEMORY_HOTPLUG is defined, before per_cpu(numa_node) is set + * up later with c_init aka intel_init/amd_init. + * So set them all (boot cpu and all APs). + */ + set_cpu_numa_node(cpu, early_cpu_to_node(cpu)); #endif #endif /* @@ -257,14 +266,6 @@ void __init setup_per_cpu_areas(void) early_per_cpu_ptr(x86_cpu_to_node_map) = NULL; #endif -#if defined(CONFIG_X86_64) && defined(CONFIG_NUMA) - /* - * make sure boot cpu numa_node is right, when boot cpu is on the - * node that doesn't have mem installed - */ - set_cpu_numa_node(boot_cpu_id, early_cpu_to_node(boot_cpu_id)); -#endif - /* Setup node to cpumask map */ setup_node_to_cpumask_map(); -- cgit v1.2.2 From a4ce96ac356e7024a7724ade9d18ba1bdf3c5c06 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Wed, 21 Jul 2010 09:25:42 -0700 Subject: Fix up trivial spelling errors ('taht' -> 'that') Pointed out by Lucas who found the new one in a comment in setup_percpu.c. And then I fixed the others that I grepped for. Reported-by: Lucas Signed-off-by: Linus Torvalds --- arch/x86/kernel/setup_percpu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index 690c2c09faf3..a60df9ae6454 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -239,7 +239,7 @@ void __init setup_per_cpu_areas(void) per_cpu(x86_cpu_to_node_map, cpu) = early_per_cpu_map(x86_cpu_to_node_map, cpu); /* - * Ensure taht the boot cpu numa_node is correct when the boot + * Ensure that the boot cpu numa_node is correct when the boot * cpu is on a node that doesn't have memory installed. * Also cpu_up() will call cpu_to_node() for APs when * MEMORY_HOTPLUG is defined, before per_cpu(numa_node) is set -- cgit v1.2.2