[PATCH v7 3/5] iommu/vt-d: Add debugfs support to show register contents
From: Gayatri KammelaDebugfs extension to dump all the register contents for each IOMMU device to the user space via debugfs. Example: root@OTC-KBLH-01:~# cat /sys/kernel/debug/intel_iommu/iommu_regset DMAR: dmar0: Register Base Address fed9 NameOffset Contents VER 0x000x0010 CAP 0x080x01cc40660462 ECAP0x100x00f0101a GCMD0x180x GSTS0x1c0xc700 RTADDR 0x200x0004071d3800 CCMD0x280x0800 FSTS0x340x FECTL 0x380x FEDATA 0x3c0xfee010044021 Cc: Fenghua Yu Cc: Jacob Pan Cc: Ashok Raj Co-Developed-by: Sohil Mehta Signed-off-by: Sohil Mehta Signed-off-by: Gayatri Kammela --- v7: Use macro for register set definitions Fix compiler warning for readq with 32bit architecture Remove leading '\n' v6: No change v5: No change v4: Fix checkpatch.pl warnings Remove error reporting for debugfs_create_file function Remove redundant IOMMU null check under for_each_active_iommu v3: Use a macro for seq file operations Change the intel_iommu_regset file name to iommu_regset Add information for MTRR registers v2: Fix seq_printf formatting drivers/iommu/intel-iommu-debug.c | 84 +++ include/linux/intel-iommu.h | 2 + 2 files changed, 86 insertions(+) diff --git a/drivers/iommu/intel-iommu-debug.c b/drivers/iommu/intel-iommu-debug.c index 8253503..38651ad 100644 --- a/drivers/iommu/intel-iommu-debug.c +++ b/drivers/iommu/intel-iommu-debug.c @@ -38,6 +38,49 @@ static const struct file_operations __name ## _fops = \ .owner = THIS_MODULE, \ } +struct iommu_regset { + int offset; + const char *regs; +}; + +#define IOMMU_REGSET_ENTRY(_reg_) \ + { DMAR_##_reg_##_REG, __stringify(_reg_) } +static const struct iommu_regset iommu_regs[] = { + IOMMU_REGSET_ENTRY(VER), + IOMMU_REGSET_ENTRY(CAP), + IOMMU_REGSET_ENTRY(ECAP), + IOMMU_REGSET_ENTRY(GCMD), + IOMMU_REGSET_ENTRY(GSTS), + IOMMU_REGSET_ENTRY(RTADDR), + IOMMU_REGSET_ENTRY(CCMD), + IOMMU_REGSET_ENTRY(FSTS), + IOMMU_REGSET_ENTRY(FECTL), + IOMMU_REGSET_ENTRY(FEDATA), + IOMMU_REGSET_ENTRY(FEADDR), + IOMMU_REGSET_ENTRY(FEUADDR), + IOMMU_REGSET_ENTRY(AFLOG), + IOMMU_REGSET_ENTRY(PMEN), + IOMMU_REGSET_ENTRY(PLMBASE), + IOMMU_REGSET_ENTRY(PLMLIMIT), + IOMMU_REGSET_ENTRY(PHMBASE), + IOMMU_REGSET_ENTRY(PHMLIMIT), + IOMMU_REGSET_ENTRY(IQH), + IOMMU_REGSET_ENTRY(IQT), + IOMMU_REGSET_ENTRY(IQA), + IOMMU_REGSET_ENTRY(ICS), + IOMMU_REGSET_ENTRY(IRTA), + IOMMU_REGSET_ENTRY(PQH), + IOMMU_REGSET_ENTRY(PQT), + IOMMU_REGSET_ENTRY(PQA), + IOMMU_REGSET_ENTRY(PRS), + IOMMU_REGSET_ENTRY(PECTL), + IOMMU_REGSET_ENTRY(PEDATA), + IOMMU_REGSET_ENTRY(PEADDR), + IOMMU_REGSET_ENTRY(PEUADDR), + IOMMU_REGSET_ENTRY(MTRRCAP), + IOMMU_REGSET_ENTRY(MTRRDEF) +}; + static void ctx_tbl_entry_show(struct seq_file *m, struct intel_iommu *iommu, int bus, bool ext) { @@ -116,6 +159,45 @@ static int dmar_translation_struct_show(struct seq_file *m, void *unused) } DEFINE_SHOW_ATTRIBUTE(dmar_translation_struct); +static int iommu_regset_show(struct seq_file *m, void *unused) +{ + struct dmar_drhd_unit *drhd; + struct intel_iommu *iommu; + unsigned long long base; + int i, ret = 0; + u64 value; + + rcu_read_lock(); + for_each_active_iommu(iommu, drhd) { + if (!drhd->reg_base_addr) { + seq_puts(m, "IOMMU: Invalid base address\n"); + ret = -EINVAL; + goto out; + } + + base = drhd->reg_base_addr; + seq_printf(m, "DMAR: %s: Register Base Address %llx\n", + iommu->name, base); + seq_puts(m, "Name\t\t\tOffset\t\tContents\n"); + /* +* Publish the contents of the 64-bit hardware registers +* by adding the offset to the pointer (virtual address). +*/ + for (i = 0 ; i < ARRAY_SIZE(iommu_regs); i++) { + value = dmar_readq(iommu->reg + iommu_regs[i].offset); + seq_printf(m, "%-8s\t\t0x%02x\t\t0x%016llx\n", + iommu_regs[i].regs,
[PATCH v7 4/5] iommu/vt-d: Add debugfs support to show Pasid table contents
From: Gayatri KammelaDebugfs extension to dump the internals such as pasid table entries for each IOMMU to the userspace. Example of such dump in Kabylake: root@OTC-KBLH-01:~# cat /sys/kernel/debug/intel_iommu/dmar_translation_struct IOMMU dmar1: Extended Root Table Address:4071d3800 Extended Root Table Entries: Bus 0 L: 4071d7001 H: 0 Lower Context Table Entries for Bus: 0 [entry] Device B:D.FLow High [16]:00:02.04071d6005 102 Higher Context Table Entries for Bus: 0 [16]:00:02.00 0 Pasid Table Address: 746cb0af Pasid Table Entries for domain 0: [Entry] Contents [0] 12c409801 Cc: Fenghua Yu Cc: Jacob Pan Cc: Ashok Raj Co-Developed-by: Sohil Mehta Signed-off-by: Sohil Mehta Signed-off-by: Gayatri Kammela --- v7: Improve code indentation and formatting v6: No change v5: No change v4: Remove the unused function parameter Fix checkpatch.pl warnings v3: No change v2: Fix seq_printf formatting drivers/iommu/intel-iommu-debug.c | 31 +++ drivers/iommu/intel-svm.c | 8 include/linux/intel-svm.h | 8 3 files changed, 39 insertions(+), 8 deletions(-) diff --git a/drivers/iommu/intel-iommu-debug.c b/drivers/iommu/intel-iommu-debug.c index 38651ad..a9a99aa 100644 --- a/drivers/iommu/intel-iommu-debug.c +++ b/drivers/iommu/intel-iommu-debug.c @@ -81,6 +81,36 @@ static const struct iommu_regset iommu_regs[] = { IOMMU_REGSET_ENTRY(MTRRDEF) }; +#ifdef CONFIG_INTEL_IOMMU_SVM +static void pasid_tbl_entry_show(struct seq_file *m, struct intel_iommu *iommu) +{ + int pasid_size = 0, i; + + if (!ecap_pasid(iommu->ecap)) + return; + + pasid_size = intel_iommu_get_pts(iommu); + seq_printf(m, "Pasid Table Address: %p\n", iommu->pasid_table); + + if (!iommu->pasid_table) + return; + + seq_printf(m, "Pasid Table Entries for domain %d:\n", iommu->segment); + seq_puts(m, "[Entry]\t\tContents\n"); + + /* Publish the pasid table entries here */ + for (i = 0; i < pasid_size; i++) { + if (!iommu->pasid_table[i].val) + continue; + + seq_printf(m, "[%d]\t\t%04llx\n", i, iommu->pasid_table[i].val); + } +} +#else /* CONFIG_INTEL_IOMMU_SVM */ +static inline void +pasid_tbl_entry_show(struct seq_file *m, struct intel_iommu *iommu) {} +#endif /* CONFIG_INTEL_IOMMU_SVM */ + static void ctx_tbl_entry_show(struct seq_file *m, struct intel_iommu *iommu, int bus, bool ext) { @@ -116,6 +146,7 @@ static void ctx_tbl_entry_show(struct seq_file *m, struct intel_iommu *iommu, iommu->segment, bus, PCI_SLOT(ctx), PCI_FUNC(ctx), context[1].lo, context[1].hi); } + pasid_tbl_entry_show(m, iommu); out: spin_unlock_irqrestore(>lock, flags); } diff --git a/drivers/iommu/intel-svm.c b/drivers/iommu/intel-svm.c index ed1cf7c..c646724 100644 --- a/drivers/iommu/intel-svm.c +++ b/drivers/iommu/intel-svm.c @@ -28,14 +28,6 @@ static irqreturn_t prq_event_thread(int irq, void *d); -struct pasid_entry { - u64 val; -}; - -struct pasid_state_entry { - u64 val; -}; - int intel_svm_alloc_pasid_tables(struct intel_iommu *iommu) { struct page *pages; diff --git a/include/linux/intel-svm.h b/include/linux/intel-svm.h index 733eaf9..a8abad6 100644 --- a/include/linux/intel-svm.h +++ b/include/linux/intel-svm.h @@ -18,6 +18,14 @@ struct device; +struct pasid_entry { + u64 val; +}; + +struct pasid_state_entry { + u64 val; +}; + struct svm_dev_ops { void (*fault_cb)(struct device *dev, int pasid, u64 address, u32 private, int rwxp, int response); -- 2.7.4 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
[PATCH v7 2/5] iommu/vt-d: Enable debugfs support to show context internals
From: Gayatri KammelaAdd a new config option CONFIG_INTEL_IOMMU_DEBUG and export Intel IOMMU internals states, such as root and context in debugfs to the userspace. Example of such dump in Kabylake: root@OTC-KBLH-01:~# cat /sys/kernel/debug/intel_iommu/dmar_translation_struct IOMMU dmar1: Extended Root Table Address:4071d3800 Extended Root Table Entries: Bus 0 L: 4071d7001 H: 0 Lower Context Table Entries for Bus: 0 [entry] Device B:D.FLow High [16]:00:02.04071d6005 102 Higher Context Table Entries for Bus: 0 [16]:00:02.00 0 IOMMU dmar0: Extended Root Table Address:4071d4800 IOMMU dmar2: Root Table Address:4071d5000 Root Table Entries: Bus 0 L: 406d13001 H: 0 Context Table Entries for Bus: 0 [entry] Device B:D.FLow High [160] :00:14.0406d12001 102 [184] :00:17.0405756001 302 [248] :00:1f.0406d3b001 202 [251] :00:1f.3405497001 402 [254] :00:1f.640662e001 502 Root Table Entries: Bus 1 L: 401e03001 H: 0 Context Table Entries for Bus: 1 [entry] Device B:D.FLow High [0] :01:00.0401e04001 602 Cc: Fenghua Yu Cc: Ashok Raj Co-Developed-by: Sohil Mehta Signed-off-by: Jacob Pan Signed-off-by: Sohil Mehta Signed-off-by: Gayatri Kammela --- v7: Split patch 1/5 and 2/5 differently Update commit message and copyright year Fix typo in a comment Simplify code v6: Change the order of includes to an alphabetical order Change seq_printf formatting v5: Change to a SPDX license tag Fix seq_printf formatting v4: Remove the unused function parameter Fix checkpatch.pl warnings Remove error reporting for debugfs_create_file function Fix unnecessary reprogramming of the context entries Simplify and merge the show context and extended context patch into one Remove redundant IOMMU null check under for_each_active_iommu v3: Add a macro for seq file operations Change the intel_iommu_ctx file name to dmar_translation_struct v2: No change drivers/iommu/Kconfig | 8 +++ drivers/iommu/Makefile| 1 + drivers/iommu/intel-iommu-debug.c | 129 ++ drivers/iommu/intel-iommu.c | 1 + include/linux/intel-iommu.h | 6 ++ 5 files changed, 145 insertions(+) create mode 100644 drivers/iommu/intel-iommu-debug.c diff --git a/drivers/iommu/Kconfig b/drivers/iommu/Kconfig index f3a2134..332648f 100644 --- a/drivers/iommu/Kconfig +++ b/drivers/iommu/Kconfig @@ -152,6 +152,14 @@ config INTEL_IOMMU and include PCI device scope covered by these DMA remapping devices. +config INTEL_IOMMU_DEBUG + bool "Export Intel IOMMU internals in Debugfs" + depends on INTEL_IOMMU && DEBUG_FS + help + Debugfs support to export IOMMU context internals, register contents, + PASID internals and interrupt remapping. To access this information in + sysfs, say Y. + config INTEL_IOMMU_SVM bool "Support for Shared Virtual Memory with Intel IOMMU" depends on INTEL_IOMMU && X86 diff --git a/drivers/iommu/Makefile b/drivers/iommu/Makefile index 1fb6958..fdbaf46 100644 --- a/drivers/iommu/Makefile +++ b/drivers/iommu/Makefile @@ -15,6 +15,7 @@ obj-$(CONFIG_ARM_SMMU) += arm-smmu.o obj-$(CONFIG_ARM_SMMU_V3) += arm-smmu-v3.o obj-$(CONFIG_DMAR_TABLE) += dmar.o obj-$(CONFIG_INTEL_IOMMU) += intel-iommu.o +obj-$(CONFIG_INTEL_IOMMU_DEBUG) += intel-iommu-debug.o obj-$(CONFIG_INTEL_IOMMU_SVM) += intel-svm.o obj-$(CONFIG_IPMMU_VMSA) += ipmmu-vmsa.o obj-$(CONFIG_IRQ_REMAP) += intel_irq_remapping.o irq_remapping.o diff --git a/drivers/iommu/intel-iommu-debug.c b/drivers/iommu/intel-iommu-debug.c new file mode 100644 index 000..8253503 --- /dev/null +++ b/drivers/iommu/intel-iommu-debug.c @@ -0,0 +1,129 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright © 2018 Intel Corporation. + * + * Authors: Gayatri Kammela + * Jacob Pan + * Sohil Mehta + */ + +#define pr_fmt(fmt) "INTEL_IOMMU: " fmt +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "irq_remapping.h" + +#define TOTAL_BUS_NR 256 /* full bus range */ +#define DEFINE_SHOW_ATTRIBUTE(__name) \ +static int __name ## _open(struct inode *inode, struct file *file) \ +{ \ + return single_open(file, __name ## _show, inode->i_private);\ +} \ +static const struct file_operations
[PATCH v7 5/5] iommu/vt-d: Add debugfs support for Interrupt remapping
Debugfs extension for Intel IOMMU to dump Interrupt remapping table entries for Interrupt remapping and Interrupt posting. The file /sys/kernel/debug/intel_iommu/ir_translation_struct provides detailed information, such as Index, Source Id, Destination Id, Vector and the IRTE values for entries with the present bit set, in the format shown. Remapped Interrupt supported on IOMMU: dmar7 IR table address:85e50 Index SrcID DstIDVct IRTE_highIRTE_low 1 f0f8 0100 30 0004f0f8 013d 7 f0f8 0400 22 0004f0f8 0422000d Posted Interrupt supported on IOMMU: dmar5 IR table address:85ec0 Index SrcID PDA_high PDA_low Vct IRTE_high IRTE_low 4 4300 000f ff765980 41 000f00044300ff76598000418001 5 4300 000f ff765980 51 000f00044300ff76598000518001 Cc: Jacob PanCc: Fenghua Yu Cc: Ashok Raj Co-Developed-by: Gayatri Kammela Signed-off-by: Gayatri Kammela Signed-off-by: Sohil Mehta --- v7: Print the IR table physical base address Simplify IR table formatting v6: Change a couple of seq_puts to seq_putc v5: Fix seq_puts formatting and remove leading '\n's v4: Remove the unused function parameter Fix checkpatch.pl warnings Remove error reporting for debugfs_create_file function Remove redundant IOMMU null check under for_each_active_iommu v3: Use a macro for seq file operations Change the intel_iommu_interrupt_remap file name to ir_translation_struct v2: Handle the case when IR is not enabled. Fix seq_printf formatting drivers/iommu/intel-iommu-debug.c | 94 +++ 1 file changed, 94 insertions(+) diff --git a/drivers/iommu/intel-iommu-debug.c b/drivers/iommu/intel-iommu-debug.c index a9a99aa..b66a073 100644 --- a/drivers/iommu/intel-iommu-debug.c +++ b/drivers/iommu/intel-iommu-debug.c @@ -229,6 +229,96 @@ static int iommu_regset_show(struct seq_file *m, void *unused) } DEFINE_SHOW_ATTRIBUTE(iommu_regset); +#ifdef CONFIG_IRQ_REMAP +static void ir_tbl_remap_entry_show(struct seq_file *m, + struct intel_iommu *iommu) +{ + struct irte *ri_entry; + int idx; + + seq_puts(m, " Index SrcID DstIDVct IRTE_high\t\tIRTE_low\n"); + + for (idx = 0; idx < INTR_REMAP_TABLE_ENTRIES; idx++) { + ri_entry = >ir_table->base[idx]; + if (!ri_entry->present || ri_entry->p_pst) + continue; + + seq_printf(m, " %d\t%04x %08x %02x %016llx\t%016llx\n", idx, + ri_entry->sid, ri_entry->dest_id, ri_entry->vector, + ri_entry->high, ri_entry->low); + } +} + +static void ir_tbl_posted_entry_show(struct seq_file *m, +struct intel_iommu *iommu) +{ + struct irte *pi_entry; + int idx; + + seq_puts(m, " Index SrcID PDA_high PDA_low Vct IRTE_high\t\tIRTE_low\n"); + + for (idx = 0; idx < INTR_REMAP_TABLE_ENTRIES; idx++) { + pi_entry = >ir_table->base[idx]; + if (!pi_entry->present || !pi_entry->p_pst) + continue; + + seq_printf(m, " %d\t%04x %08x %08x %02x %016llx\t%016llx\n", + idx, pi_entry->sid, pi_entry->pda_h, + pi_entry->pda_l << 6, pi_entry->vector, + pi_entry->high, pi_entry->low); + } +} + +/* + * For active IOMMUs go through the Interrupt remapping + * table and print valid entries in a table format for + * Remapped and Posted Interrupts. + */ +static int ir_translation_struct_show(struct seq_file *m, void *unused) +{ + struct dmar_drhd_unit *drhd; + struct intel_iommu *iommu; + u64 irta; + + rcu_read_lock(); + for_each_active_iommu(iommu, drhd) { + if (!ecap_ir_support(iommu->ecap)) + continue; + + irta = dmar_readq(iommu->reg + DMAR_IRTA_REG) & VTD_PAGE_MASK; + seq_printf(m, "Remapped Interrupt supported on IOMMU: %s\n" + " IR table address:%llx\n", iommu->name, irta); + + if (iommu->ir_table && irta) + ir_tbl_remap_entry_show(m, iommu); + else + seq_puts(m, "Interrupt Remapping is not enabled\n"); + seq_putc(m, '\n'); + } + + seq_puts(m, "\n\n"); + + for_each_active_iommu(iommu, drhd) { + if (!cap_pi_support(iommu->cap)) + continue; + + irta = dmar_readq(iommu->reg + DMAR_IRTA_REG) & VTD_PAGE_MASK; + seq_printf(m, "Posted Interrupt supported on IOMMU: %s\n" + " IR table
[PATCH v7 1/5] iommu/vt-d: Relocate struct/function declarations to its header files
From: Gayatri KammelaTo reuse the static functions and the struct declarations, move them to corresponding header files and export the needed functions. Cc: Sohil Mehta Cc: Fenghua Yu Cc: Ashok Raj Signed-off-by: Jacob Pan Signed-off-by: Gayatri Kammela --- v7: Split patch 1/5 and 2/5 differently Update the commit message v6: No change v5: No change v4: No change v3: No change v2: No change drivers/iommu/intel-iommu.c | 33 - include/linux/intel-iommu.h | 31 +++ include/linux/intel-svm.h | 2 +- 3 files changed, 36 insertions(+), 30 deletions(-) diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c index 4a2de34..f6241f6 100644 --- a/drivers/iommu/intel-iommu.c +++ b/drivers/iommu/intel-iommu.c @@ -183,16 +183,6 @@ static int rwbf_quirk; static int force_on = 0; int intel_iommu_tboot_noforce; -/* - * 0: Present - * 1-11: Reserved - * 12-63: Context Ptr (12 - (haw-1)) - * 64-127: Reserved - */ -struct root_entry { - u64 lo; - u64 hi; -}; #define ROOT_ENTRY_NR (VTD_PAGE_SIZE/sizeof(struct root_entry)) /* @@ -218,21 +208,6 @@ static phys_addr_t root_entry_uctp(struct root_entry *re) return re->hi & VTD_PAGE_MASK; } -/* - * low 64 bits: - * 0: present - * 1: fault processing disable - * 2-3: translation type - * 12-63: address space root - * high 64 bits: - * 0-2: address width - * 3-6: aval - * 8-23: domain id - */ -struct context_entry { - u64 lo; - u64 hi; -}; static inline void context_clear_pasid_enable(struct context_entry *context) { @@ -259,7 +234,7 @@ static inline bool __context_present(struct context_entry *context) return (context->lo & 1); } -static inline bool context_present(struct context_entry *context) +bool context_present(struct context_entry *context) { return context_pasid_enabled(context) ? __context_present(context) : @@ -819,8 +794,8 @@ static void domain_update_iommu_cap(struct dmar_domain *domain) domain->iommu_superpage = domain_update_iommu_superpage(NULL); } -static inline struct context_entry *iommu_context_addr(struct intel_iommu *iommu, - u8 bus, u8 devfn, int alloc) +struct context_entry *iommu_context_addr(struct intel_iommu *iommu, u8 bus, +u8 devfn, int alloc) { struct root_entry *root = >root_entry[bus]; struct context_entry *context; @@ -5208,7 +5183,7 @@ static void intel_iommu_put_resv_regions(struct device *dev, #ifdef CONFIG_INTEL_IOMMU_SVM #define MAX_NR_PASID_BITS (20) -static inline unsigned long intel_iommu_get_pts(struct intel_iommu *iommu) +unsigned long intel_iommu_get_pts(struct intel_iommu *iommu) { /* * Convert ecap_pss to extend context entry pts encoding, also diff --git a/include/linux/intel-iommu.h b/include/linux/intel-iommu.h index f3274d9..78ec85a 100644 --- a/include/linux/intel-iommu.h +++ b/include/linux/intel-iommu.h @@ -383,6 +383,33 @@ struct pasid_entry; struct pasid_state_entry; struct page_req_dsc; +/* + * 0: Present + * 1-11: Reserved + * 12-63: Context Ptr (12 - (haw-1)) + * 64-127: Reserved + */ +struct root_entry { + u64 lo; + u64 hi; +}; + +/* + * low 64 bits: + * 0: present + * 1: fault processing disable + * 2-3: translation type + * 12-63: address space root + * high 64 bits: + * 0-2: address width + * 3-6: aval + * 8-23: domain id + */ +struct context_entry { + u64 lo; + u64 hi; +}; + struct intel_iommu { void __iomem*reg; /* Pointer to hardware regs, virtual addr */ u64 reg_phys; /* physical address of hw register set */ @@ -488,8 +515,12 @@ struct intel_svm { extern int intel_iommu_enable_pasid(struct intel_iommu *iommu, struct intel_svm_dev *sdev); extern struct intel_iommu *intel_svm_device_to_iommu(struct device *dev); +extern unsigned long intel_iommu_get_pts(struct intel_iommu *iommu); #endif extern const struct attribute_group *intel_iommu_groups[]; +extern bool context_present(struct context_entry *context); +extern struct context_entry *iommu_context_addr(struct intel_iommu *iommu, + u8 bus, u8 devfn, int alloc); #endif diff --git a/include/linux/intel-svm.h b/include/linux/intel-svm.h index 99bc5b3..733eaf9 100644 --- a/include/linux/intel-svm.h +++ b/include/linux/intel-svm.h @@ -130,7 +130,7 @@ static inline int intel_svm_unbind_mm(struct device *dev, int pasid) BUG(); } -static int intel_svm_is_pasid_valid(struct device *dev, int pasid) +static inline int intel_svm_is_pasid_valid(struct device *dev, int pasid) { return -EINVAL; } -- 2.7.4 ___
[PATCH v7 0/5] Add Intel IOMMU debugfs support
Hi All, This series aims to add debugfs support for Intel IOMMU. It exposes IOMMU registers, internal context and dumps individual table entries to help debug Intel IOMMUs. The first patch does the ground work for the following patches by reorganizing some Intel IOMMU data structures. The following patches create a new Kconfig option - INTEL_IOMMU_DEBUG and add debugfs support for IOMMU context internals, register contents, PASID internals, and Interrupt remapping in that order. The information can be accessed in sysfs at '/sys/kernel/debug/intel_iommu/'. Regards, Sohil Changes since v6: - Split patch 1/5 and 2/5 differently - Simplify and improve code formatting - Use macro for register set definitions - Fix compiler warning for readq - Add Co-Developed-by tag to commit messages Changes since v5: - Change the order of includes to an alphabetical order - Change seq_printf and seq_puts formatting Changes since v4: - Change to a SPDX license tag - Fix seq_printf formatting and remove leading '\n's Changes since v3: - Remove an unused function parameter from some of the functions - Fix checkpatch.pl warnings - Remove error reporting for debugfs_create_file functions - Fix unnecessary reprogramming of the context entries - Simplify and merge the show context and extended context patch into one - Remove redundant IOMMU null check under for_each_active_iommu - Update the commit title to be consistent Changes since v2: - Added a macro for seq file operations based on recommendation by Andy Shevchenko. The marco can be moved to seq_file.h at a future point - Changed the debugfs file names to more relevant ones - Added information for MTRR registers in the regset file Changes since v1: - Fixed seq_printf formatting - Handled the case when Interrupt remapping is not enabled Gayatri Kammela (4): iommu/vt-d: Relocate struct/function declarations to its header files iommu/vt-d: Enable debugfs support to show context internals iommu/vt-d: Add debugfs support to show register contents iommu/vt-d: Add debugfs support to show Pasid table contents Sohil Mehta (1): iommu/vt-d: Add debugfs support for Interrupt remapping drivers/iommu/Kconfig | 8 + drivers/iommu/Makefile| 1 + drivers/iommu/intel-iommu-debug.c | 338 ++ drivers/iommu/intel-iommu.c | 34 +--- drivers/iommu/intel-svm.c | 8 - include/linux/intel-iommu.h | 39 + include/linux/intel-svm.h | 10 +- 7 files changed, 400 insertions(+), 38 deletions(-) create mode 100644 drivers/iommu/intel-iommu-debug.c -- 2.7.4 ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
Re: [PATCH 1/2] iommu: Fix iommu_unmap and iommu_unmap_fast return type
Hi Suravee, I love your patch! Perhaps something to improve: [auto build test WARNING on iommu/next] [also build test WARNING on v4.15 next-20180202] [if your patch is applied to the wrong git tree, please drop us a note to help improve the system] url: https://github.com/0day-ci/linux/commits/Suravee-Suthikulpanit/iommu-Fix-iommu_unmap-and-iommu_unmap_fast-return-type/20180203-015316 base: https://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu.git next reproduce: # apt-get install sparse make ARCH=x86_64 allmodconfig make C=1 CF=-D__CHECK_ENDIAN__ sparse warnings: (new ones prefixed by >>) >> drivers/iommu/qcom_iommu.c:592:27: sparse: incorrect type in initializer >> (different signedness) @@ expected long ( )( ... ) @@ got unsigned long ( )( >> ... ) @@ drivers/iommu/qcom_iommu.c:592:27: expected long ( )( ... ) drivers/iommu/qcom_iommu.c:592:27: got unsigned long ( )( ... ) drivers/iommu/qcom_iommu.c:592:12: error: initialization from incompatible pointer type .unmap = qcom_iommu_unmap, ^~~~ drivers/iommu/qcom_iommu.c:592:12: note: (near initialization for 'qcom_iommu_ops.unmap') cc1: some warnings being treated as errors vim +592 drivers/iommu/qcom_iommu.c 0ae349a0f3 Rob Clark2017-08-09 584 0ae349a0f3 Rob Clark2017-08-09 585 static const struct iommu_ops qcom_iommu_ops = { 0ae349a0f3 Rob Clark2017-08-09 586 .capable= qcom_iommu_capable, 0ae349a0f3 Rob Clark2017-08-09 587 .domain_alloc = qcom_iommu_domain_alloc, 0ae349a0f3 Rob Clark2017-08-09 588 .domain_free= qcom_iommu_domain_free, 0ae349a0f3 Rob Clark2017-08-09 589 .attach_dev = qcom_iommu_attach_dev, 0ae349a0f3 Rob Clark2017-08-09 590 .detach_dev = qcom_iommu_detach_dev, 0ae349a0f3 Rob Clark2017-08-09 591 .map= qcom_iommu_map, 0ae349a0f3 Rob Clark2017-08-09 @592 .unmap = qcom_iommu_unmap, 0ae349a0f3 Rob Clark2017-08-09 593 .map_sg = default_iommu_map_sg, 4d689b6194 Robin Murphy 2017-09-28 594 .flush_iotlb_all = qcom_iommu_iotlb_sync, 4d689b6194 Robin Murphy 2017-09-28 595 .iotlb_sync = qcom_iommu_iotlb_sync, 0ae349a0f3 Rob Clark2017-08-09 596 .iova_to_phys = qcom_iommu_iova_to_phys, 0ae349a0f3 Rob Clark2017-08-09 597 .add_device = qcom_iommu_add_device, 0ae349a0f3 Rob Clark2017-08-09 598 .remove_device = qcom_iommu_remove_device, 0ae349a0f3 Rob Clark2017-08-09 599 .device_group = generic_device_group, 0ae349a0f3 Rob Clark2017-08-09 600 .of_xlate = qcom_iommu_of_xlate, 0ae349a0f3 Rob Clark2017-08-09 601 .pgsize_bitmap = SZ_4K | SZ_64K | SZ_1M | SZ_16M, 0ae349a0f3 Rob Clark2017-08-09 602 }; 0ae349a0f3 Rob Clark2017-08-09 603 :: The code at line 592 was first introduced by commit :: 0ae349a0f33fb040a2bc228fdc6d60111455feab iommu/qcom: Add qcom_iommu :: TO: Rob Clark <robdcl...@gmail.com> :: CC: Joerg Roedel <jroe...@suse.de> --- 0-DAY kernel test infrastructureOpen Source Technology Center https://lists.01.org/pipermail/kbuild-all Intel Corporation ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
Re: [PATCH 1/2] iommu: Fix iommu_unmap and iommu_unmap_fast return type
Hi Suravee, I love your patch! Yet something to improve: [auto build test ERROR on iommu/next] [also build test ERROR on v4.15 next-20180202] [if your patch is applied to the wrong git tree, please drop us a note to help improve the system] url: https://github.com/0day-ci/linux/commits/Suravee-Suthikulpanit/iommu-Fix-iommu_unmap-and-iommu_unmap_fast-return-type/20180203-015316 base: https://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu.git next config: sparc64-allyesconfig (attached as .config) compiler: sparc64-linux-gnu-gcc (Debian 7.2.0-11) 7.2.0 reproduce: wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross # save the attached .config to linux build tree make.cross ARCH=sparc64 All errors (new ones prefixed by >>): >> drivers/iommu/qcom_iommu.c:592:12: error: initialization from incompatible >> pointer type [-Werror=incompatible-pointer-types] .unmap = qcom_iommu_unmap, ^~~~ drivers/iommu/qcom_iommu.c:592:12: note: (near initialization for 'qcom_iommu_ops.unmap') cc1: some warnings being treated as errors vim +592 drivers/iommu/qcom_iommu.c 0ae349a0f3 Rob Clark2017-08-09 584 0ae349a0f3 Rob Clark2017-08-09 585 static const struct iommu_ops qcom_iommu_ops = { 0ae349a0f3 Rob Clark2017-08-09 586 .capable= qcom_iommu_capable, 0ae349a0f3 Rob Clark2017-08-09 587 .domain_alloc = qcom_iommu_domain_alloc, 0ae349a0f3 Rob Clark2017-08-09 588 .domain_free= qcom_iommu_domain_free, 0ae349a0f3 Rob Clark2017-08-09 589 .attach_dev = qcom_iommu_attach_dev, 0ae349a0f3 Rob Clark2017-08-09 590 .detach_dev = qcom_iommu_detach_dev, 0ae349a0f3 Rob Clark2017-08-09 591 .map= qcom_iommu_map, 0ae349a0f3 Rob Clark2017-08-09 @592 .unmap = qcom_iommu_unmap, 0ae349a0f3 Rob Clark2017-08-09 593 .map_sg = default_iommu_map_sg, 4d689b6194 Robin Murphy 2017-09-28 594 .flush_iotlb_all = qcom_iommu_iotlb_sync, 4d689b6194 Robin Murphy 2017-09-28 595 .iotlb_sync = qcom_iommu_iotlb_sync, 0ae349a0f3 Rob Clark2017-08-09 596 .iova_to_phys = qcom_iommu_iova_to_phys, 0ae349a0f3 Rob Clark2017-08-09 597 .add_device = qcom_iommu_add_device, 0ae349a0f3 Rob Clark2017-08-09 598 .remove_device = qcom_iommu_remove_device, 0ae349a0f3 Rob Clark2017-08-09 599 .device_group = generic_device_group, 0ae349a0f3 Rob Clark2017-08-09 600 .of_xlate = qcom_iommu_of_xlate, 0ae349a0f3 Rob Clark2017-08-09 601 .pgsize_bitmap = SZ_4K | SZ_64K | SZ_1M | SZ_16M, 0ae349a0f3 Rob Clark2017-08-09 602 }; 0ae349a0f3 Rob Clark2017-08-09 603 :: The code at line 592 was first introduced by commit :: 0ae349a0f33fb040a2bc228fdc6d60111455feab iommu/qcom: Add qcom_iommu :: TO: Rob Clark <robdcl...@gmail.com> :: CC: Joerg Roedel <jroe...@suse.de> --- 0-DAY kernel test infrastructureOpen Source Technology Center https://lists.01.org/pipermail/kbuild-all Intel Corporation .config.gz Description: application/gzip ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
Re: [PATCH] iommu/vt-d: add NUMA awareness to intel_alloc_coherent()
On Fri, Feb 2, 2018 at 10:53 AM, Christoph Hellwigwrote: > I've got patches pending to replace all that code with > dma_direct_alloc, which will do the right thing. They were > submitted for 4.16, and I will resend them after -rc1. I see, thanks Christoph ! ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
Re: [PATCH 1/2] iommu: Fix iommu_unmap and iommu_unmap_fast return type
Hi Suravee, I love your patch! Yet something to improve: [auto build test ERROR on iommu/next] [also build test ERROR on v4.15 next-20180202] [if your patch is applied to the wrong git tree, please drop us a note to help improve the system] url: https://github.com/0day-ci/linux/commits/Suravee-Suthikulpanit/iommu-Fix-iommu_unmap-and-iommu_unmap_fast-return-type/20180203-015316 base: https://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu.git next config: s390-allyesconfig (attached as .config) compiler: s390x-linux-gnu-gcc (Debian 7.2.0-11) 7.2.0 reproduce: wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross # save the attached .config to linux build tree make.cross ARCH=s390 All errors (new ones prefixed by >>): >> drivers//iommu/s390-iommu.c:373:11: error: initialization from incompatible >> pointer type [-Werror=incompatible-pointer-types] .unmap = s390_iommu_unmap, ^~~~ drivers//iommu/s390-iommu.c:373:11: note: (near initialization for 's390_iommu_ops.unmap') cc1: some warnings being treated as errors vim +373 drivers//iommu/s390-iommu.c f42c22351 Joerg Roedel2017-04-27 365 cceb84519 Arvind Yadav2017-08-28 366 static const struct iommu_ops s390_iommu_ops = { 8128f23c4 Gerald Schaefer 2015-08-27 367 .capable = s390_iommu_capable, 8128f23c4 Gerald Schaefer 2015-08-27 368 .domain_alloc = s390_domain_alloc, 8128f23c4 Gerald Schaefer 2015-08-27 369 .domain_free = s390_domain_free, 8128f23c4 Gerald Schaefer 2015-08-27 370 .attach_dev = s390_iommu_attach_device, 8128f23c4 Gerald Schaefer 2015-08-27 371 .detach_dev = s390_iommu_detach_device, 8128f23c4 Gerald Schaefer 2015-08-27 372 .map = s390_iommu_map, 8128f23c4 Gerald Schaefer 2015-08-27 @373 .unmap = s390_iommu_unmap, 8128f23c4 Gerald Schaefer 2015-08-27 374 .iova_to_phys = s390_iommu_iova_to_phys, 8128f23c4 Gerald Schaefer 2015-08-27 375 .add_device = s390_iommu_add_device, 8128f23c4 Gerald Schaefer 2015-08-27 376 .remove_device = s390_iommu_remove_device, 0929deca4 Joerg Roedel2017-06-15 377 .device_group = generic_device_group, 8128f23c4 Gerald Schaefer 2015-08-27 378 .pgsize_bitmap = S390_IOMMU_PGSIZES, 8128f23c4 Gerald Schaefer 2015-08-27 379 }; 8128f23c4 Gerald Schaefer 2015-08-27 380 :: The code at line 373 was first introduced by commit :: 8128f23c436d0dd4f72412e1bf9256e424479dc3 iommu/s390: Add iommu api for s390 pci devices :: TO: Gerald Schaefer <gerald.schae...@de.ibm.com> :: CC: Joerg Roedel <jroe...@suse.de> --- 0-DAY kernel test infrastructureOpen Source Technology Center https://lists.01.org/pipermail/kbuild-all Intel Corporation .config.gz Description: application/gzip ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
Re: [PATCH] iommu/vt-d: add NUMA awareness to intel_alloc_coherent()
I've got patches pending to replace all that code with dma_direct_alloc, which will do the right thing. They were submitted for 4.16, and I will resend them after -rc1. ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
Re: [PATCH 22/34] dma-mapping: add an arch_dma_supported hook
On 01/12/2018 12:42 AM, Christoph Hellwig wrote: > To implement the x86 forbid_dac and iommu_sac_force we want an arch hook > so that it can apply the global options across all dma_map_ops > implementations. > > Signed-off-by: Christoph Hellwig> --- > arch/x86/include/asm/dma-mapping.h | 3 +++ > arch/x86/kernel/pci-dma.c | 19 --- > include/linux/dma-mapping.h| 11 +++ > 3 files changed, 26 insertions(+), 7 deletions(-) > diff --git a/include/linux/dma-mapping.h b/include/linux/dma-mapping.h > index 88bcb1a8211d..d67742dad904 100644 > --- a/include/linux/dma-mapping.h > +++ b/include/linux/dma-mapping.h > @@ -576,6 +576,14 @@ static inline int dma_mapping_error(struct device *dev, > dma_addr_t dma_addr) > return 0; > } > > +/* > + * This is a hack for the legacy x86 forbid_dac and iommu_sac_force. Please > + * don't use this is new code. in new code. > + */ > +#ifndef arch_dma_supported > +#define arch_dma_supported(dev, mask)(1) > +#endif -- ~Randy ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
Re: [PATCH v6 4/6] iommu/arm-smmu: Add the device_link between masters and smmu
Hi Robin, On 2/2/2018 5:01 PM, Robin Murphy wrote: > On 02/02/18 05:40, Sricharan R wrote: >> Hi Robin/Vivek, >> >> On 2/1/2018 2:23 PM, Vivek Gautam wrote: >>> Hi, >>> >>> >>> On 1/31/2018 6:39 PM, Robin Murphy wrote: On 19/01/18 11:43, Vivek Gautam wrote: > From: Sricharan R> > Finally add the device link between the master device and > smmu, so that the smmu gets runtime enabled/disabled only when the > master needs it. This is done from add_device callback which gets > called once when the master is added to the smmu. Don't we need to balance this with a device_link_del() in .remove_device (like exynos-iommu does)? >>> >>> Right. Will add device_link_del() call. Thanks for pointing out. >> >> The reason for not adding device_link_del from .remove_device was, the >> core device_del >> which calls the .remove_device from notifier, calls device_links_purge >> before that. >> That does the same thing as device_link_del. So by the time .remove_device >> is called, >> device_links for that device is already cleaned up. Vivek, you may want to >> check once that >> calling device_link_del from .remove_device has no effect, just to confirm >> once more. > > There is at least one path in which .remove_device is not called via the > notifier from device_del(), which is in the cleanup path of iommu_bus_init(). > AFAICS any links created by .add_device during that process would be left > dangling, because the device(s) would be live but otherwise disassociated > from the IOMMU afterwards. > > From a maintenance perspective it's easier to have the call in its logical > place even if it does nothing 99% of the time; that way we shouldn't have to > keep an eye out for subtle changes in the power management code or driver > core that might invalidate the device_del() reasoning above, and the power > management guys shouldn't have to comprehend the internals of the IOMMU API > to make sense of the unbalanced call if they ever want to change their API. Ha, for a moment was thinking that with probe deferral add/remove_iommu_group in iommu_bus_init is dummy. But that may not be true for all Archs. Surely agree for the maintainability reason as well. Thanks. Regards, Sricharan -- "QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
Re: [PATCH v6 4/6] iommu/arm-smmu: Add the device_link between masters and smmu
On 02/02/18 05:40, Sricharan R wrote: Hi Robin/Vivek, On 2/1/2018 2:23 PM, Vivek Gautam wrote: Hi, On 1/31/2018 6:39 PM, Robin Murphy wrote: On 19/01/18 11:43, Vivek Gautam wrote: From: Sricharan RFinally add the device link between the master device and smmu, so that the smmu gets runtime enabled/disabled only when the master needs it. This is done from add_device callback which gets called once when the master is added to the smmu. Don't we need to balance this with a device_link_del() in .remove_device (like exynos-iommu does)? Right. Will add device_link_del() call. Thanks for pointing out. The reason for not adding device_link_del from .remove_device was, the core device_del which calls the .remove_device from notifier, calls device_links_purge before that. That does the same thing as device_link_del. So by the time .remove_device is called, device_links for that device is already cleaned up. Vivek, you may want to check once that calling device_link_del from .remove_device has no effect, just to confirm once more. There is at least one path in which .remove_device is not called via the notifier from device_del(), which is in the cleanup path of iommu_bus_init(). AFAICS any links created by .add_device during that process would be left dangling, because the device(s) would be live but otherwise disassociated from the IOMMU afterwards. From a maintenance perspective it's easier to have the call in its logical place even if it does nothing 99% of the time; that way we shouldn't have to keep an eye out for subtle changes in the power management code or driver core that might invalidate the device_del() reasoning above, and the power management guys shouldn't have to comprehend the internals of the IOMMU API to make sense of the unbalanced call if they ever want to change their API. Thanks, Robin. ___ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu