[PATCH v3 1/8] x86: Kill E820_RESERVED_KERN

[PATCH v3 1/8] x86: Kill E820_RESERVED_KERN

[Yinghai Lu]

Now we are using memblock to do early resource reserver/allocation
instead of using e820 map directly, and setup_data is reserved in
memblock early already.
Also kexec generate setup_data and pass pointer to second kernel,
so second kernel reserve setup_data by their own.
(Now kexec-tools create SETUP_EFI and SETUP_E820_EXT).

We can kill E820_RESERVED_KERN and not touch e820 map at all.

That will fix bug in mark_nonsave_region that can not handle that
case: E820_RAM and E820_RESERVED_KERN ranges are continuous and
boundary is not page aligned.

Bugzilla: https://bugzilla.opensuse.org/show_bug.cgi?id=913885
Reported-by: "Lee, Chun-Yi" <jlee@suse.com>
Tested-by: "Lee, Chun-Yi" <jlee@suse.com>
Cc: "Lee, Chun-Yi" <jlee@suse.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Cc: stable@vger.kernel.org
---
 arch/x86/include/uapi/asm/e820.h |  9 ---------
 arch/x86/kernel/e820.c           |  6 ++----
 arch/x86/kernel/setup.c          | 26 --------------------------
 arch/x86/kernel/tboot.c          |  3 +--
 arch/x86/mm/init_64.c            | 11 ++++-------
 5 files changed, 7 insertions(+), 48 deletions(-)

diff --git a/arch/x86/include/uapi/asm/e820.h b/arch/x86/include/uapi/asm/e820.h
index d993e33..edc8a71 100644
--- a/arch/x86/include/uapi/asm/e820.h
+++ b/arch/x86/include/uapi/asm/e820.h
@@ -33,15 +33,6 @@
 #define E820_NVS	4
 #define E820_UNUSABLE	5
 
-
-/*
- * reserved RAM used by kernel itself
- * if CONFIG_INTEL_TXT is enabled, memory of this type will be
- * included in the S3 integrity calculation and so should not include
- * any memory that BIOS might alter over the S3 transition
- */
-#define E820_RESERVED_KERN        128
-
 #ifndef __ASSEMBLY__
 #include <linux/types.h>
 struct e820entry {
diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c
index 46201de..2a6bed9 100644
--- a/arch/x86/kernel/e820.c
+++ b/arch/x86/kernel/e820.c
@@ -134,7 +134,6 @@ static void __init e820_print_type(u32 type)
 {
 	switch (type) {
 	case E820_RAM:
-	case E820_RESERVED_KERN:
 		printk(KERN_CONT "usable");
 		break;
 	case E820_RESERVED:
@@ -688,7 +687,7 @@ void __init e820_mark_nosave_regions(unsigned long limit_pfn)
 			register_nosave_region(pfn, PFN_UP(ei->addr));
 
 		pfn = PFN_DOWN(ei->addr + ei->size);
-		if (ei->type != E820_RAM && ei->type != E820_RESERVED_KERN)
+		if (ei->type != E820_RAM)
 			register_nosave_region(PFN_UP(ei->addr), pfn);
 
 		if (pfn >= limit_pfn)
@@ -902,7 +901,6 @@ void __init finish_e820_parsing(void)
 static inline const char *e820_type_to_string(int e820_type)
 {
 	switch (e820_type) {
-	case E820_RESERVED_KERN:
 	case E820_RAM:	return "System RAM";
 	case E820_ACPI:	return "ACPI Tables";
 	case E820_NVS:	return "ACPI Non-volatile Storage";
@@ -1077,7 +1075,7 @@ void __init memblock_x86_fill(void)
 		if (end != (resource_size_t)end)
 			continue;
 
-		if (ei->type != E820_RAM && ei->type != E820_RESERVED_KERN)
+		if (ei->type != E820_RAM)
 			continue;
 
 		memblock_add(ei->addr, ei->size);
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index 912f124..aed343b 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -478,30 +478,6 @@ static void __init parse_setup_data(void)
 	}
 }
 
-static void __init e820_reserve_setup_data(void)
-{
-	struct setup_data *data;
-	u64 pa_data;
-	int found = 0;
-
-	pa_data = boot_params.hdr.setup_data;
-	while (pa_data) {
-		data = early_memremap(pa_data, sizeof(*data));
-		e820_update_range(pa_data, sizeof(*data)+data->len,
-			 E820_RAM, E820_RESERVED_KERN);
-		found = 1;
-		pa_data = data->next;
-		early_iounmap(data, sizeof(*data));
-	}
-	if (!found)
-		return;
-
-	sanitize_e820_map(e820.map, ARRAY_SIZE(e820.map), &e820.nr_map);
-	memcpy(&e820_saved, &e820, sizeof(struct e820map));
-	printk(KERN_INFO "extended physical RAM map:\n");
-	e820_print_map("reserve setup_data");
-}
-
 static void __init memblock_x86_reserve_range_setup_data(void)
 {
 	struct setup_data *data;
@@ -1037,8 +1013,6 @@ void __init setup_arch(char **cmdline_p)
 		early_dump_pci_devices();
 #endif
 
-	/* update the e820_saved too */
-	e820_reserve_setup_data();
 	finish_e820_parsing();
 
 	if (efi_enabled(EFI_BOOT))
diff --git a/arch/x86/kernel/tboot.c b/arch/x86/kernel/tboot.c
index 91a4496..3c2752a 100644
--- a/arch/x86/kernel/tboot.c
+++ b/arch/x86/kernel/tboot.c
@@ -195,8 +195,7 @@ static int tboot_setup_sleep(void)
 	tboot->num_mac_regions = 0;
 
 	for (i = 0; i < e820.nr_map; i++) {
-		if ((e820.map[i].type != E820_RAM)
-		 && (e820.map[i].type != E820_RESERVED_KERN))
+		if (e820.map[i].type != E820_RAM)
 			continue;
 
 		add_mac_region(e820.map[i].addr, e820.map[i].size);
diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
index c30efb6..63520ec 100644
--- a/arch/x86/mm/init_64.c
+++ b/arch/x86/mm/init_64.c
@@ -354,8 +354,7 @@ phys_pte_init(pte_t *pte_page, unsigned long addr, unsigned long end,
 		next = (addr & PAGE_MASK) + PAGE_SIZE;
 		if (addr >= end) {
 			if (!after_bootmem &&
-			    !e820_any_mapped(addr & PAGE_MASK, next, E820_RAM) &&
-			    !e820_any_mapped(addr & PAGE_MASK, next, E820_RESERVED_KERN))
+			    !e820_any_mapped(addr & PAGE_MASK, next, E820_RAM))
 				set_pte(pte, __pte(0));
 			continue;
 		}
@@ -401,9 +400,8 @@ phys_pmd_init(pmd_t *pmd_page, unsigned long address, unsigned long end,
 
 		next = (address & PMD_MASK) + PMD_SIZE;
 		if (address >= end) {
-			if (!after_bootmem &&
-			    !e820_any_mapped(address & PMD_MASK, next, E820_RAM) &&
-			    !e820_any_mapped(address & PMD_MASK, next, E820_RESERVED_KERN))
+			if (!after_bootmem && !e820_any_mapped(
+					address & PMD_MASK, next, E820_RAM))
 				set_pmd(pmd, __pmd(0));
 			continue;
 		}
@@ -476,8 +474,7 @@ phys_pud_init(pud_t *pud_page, unsigned long addr, unsigned long end,
 		next = (addr & PUD_MASK) + PUD_SIZE;
 		if (addr >= end) {
 			if (!after_bootmem &&
-			    !e820_any_mapped(addr & PUD_MASK, next, E820_RAM) &&
-			    !e820_any_mapped(addr & PUD_MASK, next, E820_RESERVED_KERN))
+			    !e820_any_mapped(addr & PUD_MASK, next, E820_RAM))
 				set_pud(pud, __pud(0));
 			continue;
 		}
-- 
1.8.4.5

[PATCH v3 0/8] x86, boot: clean up setup_data handling

[Yinghai Lu]

Now setup_data is reserved via memblock and e820 and different
handlers have different ways, and it is confusing.
1. SETUP_E820_EXT: is consumed early and will not copy or access again.
        have memory wasted.
2. SETUP_EFI: is accessed via ioremap every time at early stage.
        have memory wasted.
3. SETUP_DTB: is copied locally.
        have memory wasted.
4. SETUP_PCI: is accessed via ioremap for every pci devices, even run-time.
5. SETUP_KASLR: is accessed early, will not copy or access again.
        have memory wasted.

Also setup_data is exported to debugfs for debug purpose.

Here will convert to let every handler to decide how to handle it.
and will not reserve the setup_data generally, so will not
waste memory and also make memblock/e820 keep page aligned.
1. not touch E820 anymore.
2. copy SETUP_EFI to __initdata variable and access it without ioremap.
3. SETUP_DTB: reserver and copy to local and free.
4. SETUP_PCI: reverve localy and convert to list, to avoid keeping ioremap.
5. export SETUP_PCI via sysfs.

Also put them in:
git://git.kernel.org/pub/scm/linux/kernel/git/yinghai/linux-yinghai.git for-x86-4.0-rc2-setup_data

Should be materials for v4.1.

Thanks

Yinghai Lu

-v3: separated with kasl patches, and fix early_ioremap return checking.

Yinghai Lu (8):
  x86: Kill E820_RESERVED_KERN
  x86, efi: Copy SETUP_EFI data and access directly
  x86, of: Let add_dtb reserve setup_data locally
  x86, boot: Add add_pci handler for SETUP_PCI
  x86: Kill not used setup_data handling code
  x86, boot, PCI: Convert SETUP_PCI data to list
  x86, boot, PCI: Copy SETUP_PCI rom to kernel space
  x86, boot, PCI: Export SETUP_PCI data via sysfs

 arch/x86/include/asm/efi.h       |   2 +-
 arch/x86/include/asm/pci.h       |   4 +
 arch/x86/include/asm/prom.h      |   9 +-
 arch/x86/include/uapi/asm/e820.h |   9 --
 arch/x86/kernel/devicetree.c     |  43 +++---
 arch/x86/kernel/e820.c           |   6 +-
 arch/x86/kernel/kdebugfs.c       | 142 ------------------
 arch/x86/kernel/setup.c          |  52 ++-----
 arch/x86/kernel/tboot.c          |   3 +-
 arch/x86/mm/init_64.c            |  11 +-
 arch/x86/pci/common.c            | 316 ++++++++++++++++++++++++++++++++++++---
 arch/x86/platform/efi/efi.c      |  13 +-
 arch/x86/platform/efi/efi_64.c   |  13 +-
 arch/x86/platform/efi/quirks.c   |  23 +--
 14 files changed, 371 insertions(+), 275 deletions(-)

-- 
1.8.4.5

[PATCH v3 1/8] x86: Kill E820_RESERVED_KERN

[Yinghai Lu]

Now we are using memblock to do early resource reserver/allocation
instead of using e820 map directly, and setup_data is reserved in
memblock early already.
Also kexec generate setup_data and pass pointer to second kernel,
so second kernel reserve setup_data by their own.
(Now kexec-tools create SETUP_EFI and SETUP_E820_EXT).

We can kill E820_RESERVED_KERN and not touch e820 map at all.

That will fix bug in mark_nonsave_region that can not handle that
case: E820_RAM and E820_RESERVED_KERN ranges are continuous and
boundary is not page aligned.

Bugzilla: https://bugzilla.opensuse.org/show_bug.cgi?id=913885
Reported-by: "Lee, Chun-Yi" <jlee@suse.com>
Tested-by: "Lee, Chun-Yi" <jlee@suse.com>
Cc: "Lee, Chun-Yi" <jlee@suse.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Cc: stable@vger.kernel.org
---
 arch/x86/include/uapi/asm/e820.h |  9 ---------
 arch/x86/kernel/e820.c           |  6 ++----
 arch/x86/kernel/setup.c          | 26 --------------------------
 arch/x86/kernel/tboot.c          |  3 +--
 arch/x86/mm/init_64.c            | 11 ++++-------
 5 files changed, 7 insertions(+), 48 deletions(-)

diff --git a/arch/x86/include/uapi/asm/e820.h b/arch/x86/include/uapi/asm/e820.h
index d993e33..edc8a71 100644
--- a/arch/x86/include/uapi/asm/e820.h
+++ b/arch/x86/include/uapi/asm/e820.h
@@ -33,15 +33,6 @@
 #define E820_NVS	4
 #define E820_UNUSABLE	5
 
-
-/*
- * reserved RAM used by kernel itself
- * if CONFIG_INTEL_TXT is enabled, memory of this type will be
- * included in the S3 integrity calculation and so should not include
- * any memory that BIOS might alter over the S3 transition
- */
-#define E820_RESERVED_KERN        128
-
 #ifndef __ASSEMBLY__
 #include <linux/types.h>
 struct e820entry {
diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c
index 46201de..2a6bed9 100644
--- a/arch/x86/kernel/e820.c
+++ b/arch/x86/kernel/e820.c
@@ -134,7 +134,6 @@ static void __init e820_print_type(u32 type)
 {
 	switch (type) {
 	case E820_RAM:
-	case E820_RESERVED_KERN:
 		printk(KERN_CONT "usable");
 		break;
 	case E820_RESERVED:
@@ -688,7 +687,7 @@ void __init e820_mark_nosave_regions(unsigned long limit_pfn)
 			register_nosave_region(pfn, PFN_UP(ei->addr));
 
 		pfn = PFN_DOWN(ei->addr + ei->size);
-		if (ei->type != E820_RAM && ei->type != E820_RESERVED_KERN)
+		if (ei->type != E820_RAM)
 			register_nosave_region(PFN_UP(ei->addr), pfn);
 
 		if (pfn >= limit_pfn)
@@ -902,7 +901,6 @@ void __init finish_e820_parsing(void)
 static inline const char *e820_type_to_string(int e820_type)
 {
 	switch (e820_type) {
-	case E820_RESERVED_KERN:
 	case E820_RAM:	return "System RAM";
 	case E820_ACPI:	return "ACPI Tables";
 	case E820_NVS:	return "ACPI Non-volatile Storage";
@@ -1077,7 +1075,7 @@ void __init memblock_x86_fill(void)
 		if (end != (resource_size_t)end)
 			continue;
 
-		if (ei->type != E820_RAM && ei->type != E820_RESERVED_KERN)
+		if (ei->type != E820_RAM)
 			continue;
 
 		memblock_add(ei->addr, ei->size);
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index 912f124..aed343b 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -478,30 +478,6 @@ static void __init parse_setup_data(void)
 	}
 }
 
-static void __init e820_reserve_setup_data(void)
-{
-	struct setup_data *data;
-	u64 pa_data;
-	int found = 0;
-
-	pa_data = boot_params.hdr.setup_data;
-	while (pa_data) {
-		data = early_memremap(pa_data, sizeof(*data));
-		e820_update_range(pa_data, sizeof(*data)+data->len,
-			 E820_RAM, E820_RESERVED_KERN);
-		found = 1;
-		pa_data = data->next;
-		early_iounmap(data, sizeof(*data));
-	}
-	if (!found)
-		return;
-
-	sanitize_e820_map(e820.map, ARRAY_SIZE(e820.map), &e820.nr_map);
-	memcpy(&e820_saved, &e820, sizeof(struct e820map));
-	printk(KERN_INFO "extended physical RAM map:\n");
-	e820_print_map("reserve setup_data");
-}
-
 static void __init memblock_x86_reserve_range_setup_data(void)
 {
 	struct setup_data *data;
@@ -1037,8 +1013,6 @@ void __init setup_arch(char **cmdline_p)
 		early_dump_pci_devices();
 #endif
 
-	/* update the e820_saved too */
-	e820_reserve_setup_data();
 	finish_e820_parsing();
 
 	if (efi_enabled(EFI_BOOT))
diff --git a/arch/x86/kernel/tboot.c b/arch/x86/kernel/tboot.c
index 91a4496..3c2752a 100644
--- a/arch/x86/kernel/tboot.c
+++ b/arch/x86/kernel/tboot.c
@@ -195,8 +195,7 @@ static int tboot_setup_sleep(void)
 	tboot->num_mac_regions = 0;
 
 	for (i = 0; i < e820.nr_map; i++) {
-		if ((e820.map[i].type != E820_RAM)
-		 && (e820.map[i].type != E820_RESERVED_KERN))
+		if (e820.map[i].type != E820_RAM)
 			continue;
 
 		add_mac_region(e820.map[i].addr, e820.map[i].size);
diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
index c30efb6..63520ec 100644
--- a/arch/x86/mm/init_64.c
+++ b/arch/x86/mm/init_64.c
@@ -354,8 +354,7 @@ phys_pte_init(pte_t *pte_page, unsigned long addr, unsigned long end,
 		next = (addr & PAGE_MASK) + PAGE_SIZE;
 		if (addr >= end) {
 			if (!after_bootmem &&
-			    !e820_any_mapped(addr & PAGE_MASK, next, E820_RAM) &&
-			    !e820_any_mapped(addr & PAGE_MASK, next, E820_RESERVED_KERN))
+			    !e820_any_mapped(addr & PAGE_MASK, next, E820_RAM))
 				set_pte(pte, __pte(0));
 			continue;
 		}
@@ -401,9 +400,8 @@ phys_pmd_init(pmd_t *pmd_page, unsigned long address, unsigned long end,
 
 		next = (address & PMD_MASK) + PMD_SIZE;
 		if (address >= end) {
-			if (!after_bootmem &&
-			    !e820_any_mapped(address & PMD_MASK, next, E820_RAM) &&
-			    !e820_any_mapped(address & PMD_MASK, next, E820_RESERVED_KERN))
+			if (!after_bootmem && !e820_any_mapped(
+					address & PMD_MASK, next, E820_RAM))
 				set_pmd(pmd, __pmd(0));
 			continue;
 		}
@@ -476,8 +474,7 @@ phys_pud_init(pud_t *pud_page, unsigned long addr, unsigned long end,
 		next = (addr & PUD_MASK) + PUD_SIZE;
 		if (addr >= end) {
 			if (!after_bootmem &&
-			    !e820_any_mapped(addr & PUD_MASK, next, E820_RAM) &&
-			    !e820_any_mapped(addr & PUD_MASK, next, E820_RESERVED_KERN))
+			    !e820_any_mapped(addr & PUD_MASK, next, E820_RAM))
 				set_pud(pud, __pud(0));
 			continue;
 		}
-- 
1.8.4.5

[PATCH v3 2/8] x86, efi: Copy SETUP_EFI data and access directly

[Yinghai Lu]

The copy will be in __initdata, and it is small.

We can use pointer to access the setup_data instead of using early_memmap
everywhere.

Cc: Matt Fleming <matt.fleming@intel.com>
Cc: linux-efi@vger.kernel.org
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
---
 arch/x86/include/asm/efi.h     |  2 +-
 arch/x86/platform/efi/efi.c    | 13 ++-----------
 arch/x86/platform/efi/efi_64.c | 13 ++++++++++++-
 arch/x86/platform/efi/quirks.c | 23 ++++++-----------------
 4 files changed, 21 insertions(+), 30 deletions(-)

diff --git a/arch/x86/include/asm/efi.h b/arch/x86/include/asm/efi.h
index 25bce45..edbecd6 100644
--- a/arch/x86/include/asm/efi.h
+++ b/arch/x86/include/asm/efi.h
@@ -114,7 +114,7 @@ struct efi_setup_data {
 	u64 reserved[8];
 };
 
-extern u64 efi_setup;
+extern struct efi_setup_data *efi_setup;
 
 #ifdef CONFIG_EFI
 
diff --git a/arch/x86/platform/efi/efi.c b/arch/x86/platform/efi/efi.c
index dbc8627..1cd38e8 100644
--- a/arch/x86/platform/efi/efi.c
+++ b/arch/x86/platform/efi/efi.c
@@ -68,7 +68,7 @@ static efi_config_table_type_t arch_tables[] __initdata = {
 	{NULL_GUID, NULL, NULL},
 };
 
-u64 efi_setup;		/* efi setup_data physical address */
+struct efi_setup_data *efi_setup __initdata; /* cached efi setup_data pointer */
 
 static int add_efi_memmap __initdata;
 static int __init setup_add_efi_memmap(char *arg)
@@ -225,20 +225,13 @@ static int __init efi_systab_init(void *phys)
 {
 	if (efi_enabled(EFI_64BIT)) {
 		efi_system_table_64_t *systab64;
-		struct efi_setup_data *data = NULL;
+		struct efi_setup_data *data = efi_setup;
 		u64 tmp = 0;
 
-		if (efi_setup) {
-			data = early_memremap(efi_setup, sizeof(*data));
-			if (!data)
-				return -ENOMEM;
-		}
 		systab64 = early_memremap((unsigned long)phys,
 					 sizeof(*systab64));
 		if (systab64 == NULL) {
 			pr_err("Couldn't map the system table!\n");
-			if (data)
-				early_memunmap(data, sizeof(*data));
 			return -ENOMEM;
 		}
 
@@ -271,8 +264,6 @@ static int __init efi_systab_init(void *phys)
 		tmp |= data ? data->tables : systab64->tables;
 
 		early_memunmap(systab64, sizeof(*systab64));
-		if (data)
-			early_memunmap(data, sizeof(*data));
 #ifdef CONFIG_X86_32
 		if (tmp >> 32) {
 			pr_err("EFI data located above 4GB, disabling EFI.\n");
diff --git a/arch/x86/platform/efi/efi_64.c b/arch/x86/platform/efi/efi_64.c
index 17e80d8..a541c6c 100644
--- a/arch/x86/platform/efi/efi_64.c
+++ b/arch/x86/platform/efi/efi_64.c
@@ -292,9 +292,20 @@ void __iomem *__init efi_ioremap(unsigned long phys_addr, unsigned long size,
 	return (void __iomem *)__va(phys_addr);
 }
 
+static struct efi_setup_data efi_setup_data __initdata;
+
 void __init parse_efi_setup(u64 phys_addr, u32 data_len)
 {
-	efi_setup = phys_addr + sizeof(struct setup_data);
+	struct efi_setup_data *data;
+
+	data = early_memremap(phys_addr + sizeof(struct setup_data),
+			      sizeof(*data));
+	if (!data)
+		return;
+
+	efi_setup_data = *data;
+	early_memunmap(data, sizeof(*data));
+	efi_setup = &efi_setup_data;
 }
 
 void __init efi_runtime_mkexec(void)
diff --git a/arch/x86/platform/efi/quirks.c b/arch/x86/platform/efi/quirks.c
index 1c7380d..45fec7d 100644
--- a/arch/x86/platform/efi/quirks.c
+++ b/arch/x86/platform/efi/quirks.c
@@ -203,9 +203,8 @@ void __init efi_free_boot_services(void)
  */
 int __init efi_reuse_config(u64 tables, int nr_tables)
 {
-	int i, sz, ret = 0;
+	int i, sz;
 	void *p, *tablep;
-	struct efi_setup_data *data;
 
 	if (!efi_setup)
 		return 0;
@@ -213,22 +212,15 @@ int __init efi_reuse_config(u64 tables, int nr_tables)
 	if (!efi_enabled(EFI_64BIT))
 		return 0;
 
-	data = early_memremap(efi_setup, sizeof(*data));
-	if (!data) {
-		ret = -ENOMEM;
-		goto out;
-	}
-
-	if (!data->smbios)
-		goto out_memremap;
+	if (!efi_setup->smbios)
+		return 0;
 
 	sz = sizeof(efi_config_table_64_t);
 
 	p = tablep = early_memremap(tables, nr_tables * sz);
 	if (!p) {
 		pr_err("Could not map Configuration table!\n");
-		ret = -ENOMEM;
-		goto out_memremap;
+		return -ENOMEM;
 	}
 
 	for (i = 0; i < efi.systab->nr_tables; i++) {
@@ -237,15 +229,12 @@ int __init efi_reuse_config(u64 tables, int nr_tables)
 		guid = ((efi_config_table_64_t *)p)->guid;
 
 		if (!efi_guidcmp(guid, SMBIOS_TABLE_GUID))
-			((efi_config_table_64_t *)p)->table = data->smbios;
+			((efi_config_table_64_t *)p)->table = efi_setup->smbios;
 		p += sz;
 	}
 	early_memunmap(tablep, nr_tables * sz);
 
-out_memremap:
-	early_memunmap(data, sizeof(*data));
-out:
-	return ret;
+	return 0;
 }
 
 void __init efi_apply_memmap_quirks(void)
-- 
1.8.4.5

[PATCH v3 3/8] x86, of: Let add_dtb reserve setup_data locally

[Yinghai Lu]

We will not reserve setup_data in generic code. Every handler need to
reserve and copy setup_data locally.

Current dtd handling already have code for copying, just add reserve code.

Also simplify code a bit by storing real dtb size.

Cc: Rob Herring <robh@kernel.org>
Cc: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
---
 arch/x86/include/asm/prom.h  |  9 ++++++---
 arch/x86/kernel/devicetree.c | 43 +++++++++++++++++++++++++------------------
 2 files changed, 31 insertions(+), 21 deletions(-)

diff --git a/arch/x86/include/asm/prom.h b/arch/x86/include/asm/prom.h
index 1d081ac..fb716eddc 100644
--- a/arch/x86/include/asm/prom.h
+++ b/arch/x86/include/asm/prom.h
@@ -24,17 +24,20 @@
 
 #ifdef CONFIG_OF
 extern int of_ioapic;
-extern u64 initial_dtb;
-extern void add_dtb(u64 data);
 void x86_of_pci_init(void);
 void x86_dtb_init(void);
 #else
-static inline void add_dtb(u64 data) { }
 static inline void x86_of_pci_init(void) { }
 static inline void x86_dtb_init(void) { }
 #define of_ioapic 0
 #endif
 
+#ifdef CONFIG_OF_FLATTREE
+extern void add_dtb(u64 data);
+#else
+static inline void add_dtb(u64 data) { }
+#endif
+
 extern char cmd_line[COMMAND_LINE_SIZE];
 
 #endif /* __ASSEMBLY__ */
diff --git a/arch/x86/kernel/devicetree.c b/arch/x86/kernel/devicetree.c
index 3d35033..55cf76a 100644
--- a/arch/x86/kernel/devicetree.c
+++ b/arch/x86/kernel/devicetree.c
@@ -2,6 +2,7 @@
  * Architecture specific OF callbacks.
  */
 #include <linux/bootmem.h>
+#include <linux/memblock.h>
 #include <linux/export.h>
 #include <linux/io.h>
 #include <linux/irqdomain.h>
@@ -23,7 +24,6 @@
 #include <asm/setup.h>
 #include <asm/i8259.h>
 
-__initdata u64 initial_dtb;
 char __initdata cmd_line[COMMAND_LINE_SIZE];
 
 int __initdata of_ioapic;
@@ -43,11 +43,27 @@ void * __init early_init_dt_alloc_memory_arch(u64 size, u64 align)
 	return __alloc_bootmem(size, align, __pa(MAX_DMA_ADDRESS));
 }
 
+#ifdef CONFIG_OF_FLATTREE
+static u64 initial_dtb __initdata;
+static u32 initial_dtb_size __initdata;
 void __init add_dtb(u64 data)
 {
+	u32 map_len;
+
 	initial_dtb = data + offsetof(struct setup_data, data);
-}
 
+	map_len = max(PAGE_SIZE - (initial_dtb & ~PAGE_MASK), (u64)128);
+	initial_boot_params = early_memremap(initial_dtb, map_len);
+	if (!initial_boot_params) {
+		initial_dtb = 0;
+		return;
+	}
+	initial_dtb_size = of_get_flat_dt_size();
+	early_memunmap(initial_boot_params, map_len);
+	initial_boot_params = NULL;
+	memblock_reserve(initial_dtb, initial_dtb_size);
+}
+#endif
 /*
  * CE4100 ids. Will be moved to machine_device_initcall() once we have it.
  */
@@ -272,31 +288,22 @@ static void __init dtb_apic_setup(void)
 	dtb_ioapic_setup();
 }
 
-#ifdef CONFIG_OF_FLATTREE
 static void __init x86_flattree_get_config(void)
 {
-	u32 size, map_len;
+#ifdef CONFIG_OF_FLATTREE
 	void *dt;
 
 	if (!initial_dtb)
 		return;
 
-	map_len = max(PAGE_SIZE - (initial_dtb & ~PAGE_MASK), (u64)128);
-
-	initial_boot_params = dt = early_memremap(initial_dtb, map_len);
-	size = of_get_flat_dt_size();
-	if (map_len < size) {
-		early_iounmap(dt, map_len);
-		initial_boot_params = dt = early_memremap(initial_dtb, size);
-		map_len = size;
-	}
-
+	initial_boot_params = dt = early_memremap(initial_dtb,
+						  initial_dtb_size);
 	unflatten_and_copy_device_tree();
-	early_iounmap(dt, map_len);
-}
-#else
-static inline void x86_flattree_get_config(void) { }
+	early_memunmap(dt, initial_dtb_size);
+
+	memblock_free(initial_dtb, initial_dtb_size);
 #endif
+}
 
 void __init x86_dtb_init(void)
 {
-- 
1.8.4.5

[PATCH v3 4/8] x86, boot: Add add_pci handler for SETUP_PCI

[Yinghai Lu]

Let it reserve setup_data, and keep it's own list.

Also clear the hdr.setup_data, as all handler now handle or
reserve setup_data locally already.

Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Matt Fleming <matt.fleming@intel.com>
Cc: linux-pci@vger.kernel.org
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
---
 arch/x86/include/asm/pci.h |  2 ++
 arch/x86/kernel/setup.c    |  8 ++++++++
 arch/x86/pci/common.c      | 45 +++++++++++++++++++++++++++++++--------------
 3 files changed, 41 insertions(+), 14 deletions(-)

diff --git a/arch/x86/include/asm/pci.h b/arch/x86/include/asm/pci.h
index 4e370a5..7fbd5f3 100644
--- a/arch/x86/include/asm/pci.h
+++ b/arch/x86/include/asm/pci.h
@@ -87,8 +87,10 @@ static inline void pci_dma_burst_advice(struct pci_dev *pdev,
 	*strat = PCI_DMA_BURST_INFINITY;
 	*strategy_parameter = ~0UL;
 }
+void add_pci(u64 pa_data);
 #else
 static inline void early_quirks(void) { }
+static inline void add_pci(u64 pa_data) { }
 #endif
 
 extern void pci_iommu_alloc(void);
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index aed343b..94f95e0 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -458,6 +458,8 @@ static void __init parse_setup_data(void)
 		pa_next = data->next;
 		early_iounmap(data, sizeof(*data));
 
+		printk(KERN_DEBUG "setup_data type: %d @ %#010llx\n",
+				data_type, pa_data);
 		switch (data_type) {
 		case SETUP_E820_EXT:
 			parse_e820_ext(pa_data, data_len);
@@ -465,6 +467,9 @@ static void __init parse_setup_data(void)
 		case SETUP_DTB:
 			add_dtb(pa_data);
 			break;
+		case SETUP_PCI:
+			add_pci(pa_data);
+			break;
 		case SETUP_EFI:
 			parse_efi_setup(pa_data, data_len);
 			break;
@@ -472,10 +477,13 @@ static void __init parse_setup_data(void)
 			parse_kaslr_setup(pa_data, data_len);
 			break;
 		default:
+			pr_warn("Unknown setup_data type: %d @ %#010llx ignored!\n",
+				data_type, pa_data);
 			break;
 		}
 		pa_data = pa_next;
 	}
+	boot_params.hdr.setup_data = 0; /* all done */
 }
 
 static void __init memblock_x86_reserve_range_setup_data(void)
diff --git a/arch/x86/pci/common.c b/arch/x86/pci/common.c
index 3d2612b..f40df5c 100644
--- a/arch/x86/pci/common.c
+++ b/arch/x86/pci/common.c
@@ -9,6 +9,7 @@
 #include <linux/pci-acpi.h>
 #include <linux/ioport.h>
 #include <linux/init.h>
+#include <linux/memblock.h>
 #include <linux/dmi.h>
 #include <linux/slab.h>
 
@@ -667,31 +668,47 @@ unsigned int pcibios_assign_all_busses(void)
 	return (pci_probe & PCI_ASSIGN_ALL_BUSSES) ? 1 : 0;
 }
 
+static u64 pci_setup_data;
+void __init add_pci(u64 pa_data)
+{
+	struct setup_data *data;
+
+	data = early_memremap(pa_data, sizeof(*data));
+	if (!data)
+		return;
+
+	memblock_reserve(pa_data, sizeof(*data) + data->len);
+	data->next = pci_setup_data;
+	pci_setup_data = pa_data;
+	early_memunmap(data, sizeof(*data));
+}
+
 int pcibios_add_device(struct pci_dev *dev)
 {
 	struct setup_data *data;
 	struct pci_setup_rom *rom;
 	u64 pa_data;
 
-	pa_data = boot_params.hdr.setup_data;
+	pa_data = pci_setup_data;
 	while (pa_data) {
 		data = ioremap(pa_data, sizeof(*rom));
 		if (!data)
 			return -ENOMEM;
 
-		if (data->type == SETUP_PCI) {
-			rom = (struct pci_setup_rom *)data;
-
-			if ((pci_domain_nr(dev->bus) == rom->segment) &&
-			    (dev->bus->number == rom->bus) &&
-			    (PCI_SLOT(dev->devfn) == rom->device) &&
-			    (PCI_FUNC(dev->devfn) == rom->function) &&
-			    (dev->vendor == rom->vendor) &&
-			    (dev->device == rom->devid)) {
-				dev->rom = pa_data +
-				      offsetof(struct pci_setup_rom, romdata);
-				dev->romlen = rom->pcilen;
-			}
+		rom = (struct pci_setup_rom *)data;
+
+		if ((pci_domain_nr(dev->bus) == rom->segment) &&
+		    (dev->bus->number == rom->bus) &&
+		    (PCI_SLOT(dev->devfn) == rom->device) &&
+		    (PCI_FUNC(dev->devfn) == rom->function) &&
+		    (dev->vendor == rom->vendor) &&
+		    (dev->device == rom->devid)) {
+			dev->rom = pa_data +
+			      offsetof(struct pci_setup_rom, romdata);
+			dev->romlen = rom->pcilen;
+			dev_printk(KERN_DEBUG, &dev->dev, "set rom to [%#010lx, %#010lx] via SETUP_PCI\n",
+				   (unsigned long)dev->rom,
+				   (unsigned long)(dev->rom + dev->romlen - 1));
 		}
 		pa_data = data->next;
 		iounmap(data);
-- 
1.8.4.5

[PATCH v3 5/8] x86: Kill not used setup_data handling code

[Yinghai Lu]

Cc: Matt Fleming <matt.fleming@intel.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
---
 arch/x86/kernel/kdebugfs.c | 142 ---------------------------------------------
 arch/x86/kernel/setup.c    |  17 ------
 2 files changed, 159 deletions(-)

diff --git a/arch/x86/kernel/kdebugfs.c b/arch/x86/kernel/kdebugfs.c
index dc1404b..c8ca86c 100644
--- a/arch/x86/kernel/kdebugfs.c
+++ b/arch/x86/kernel/kdebugfs.c
@@ -21,142 +21,6 @@ struct dentry *arch_debugfs_dir;
 EXPORT_SYMBOL(arch_debugfs_dir);
 
 #ifdef CONFIG_DEBUG_BOOT_PARAMS
-struct setup_data_node {
-	u64 paddr;
-	u32 type;
-	u32 len;
-};
-
-static ssize_t setup_data_read(struct file *file, char __user *user_buf,
-			       size_t count, loff_t *ppos)
-{
-	struct setup_data_node *node = file->private_data;
-	unsigned long remain;
-	loff_t pos = *ppos;
-	struct page *pg;
-	void *p;
-	u64 pa;
-
-	if (pos < 0)
-		return -EINVAL;
-
-	if (pos >= node->len)
-		return 0;
-
-	if (count > node->len - pos)
-		count = node->len - pos;
-
-	pa = node->paddr + sizeof(struct setup_data) + pos;
-	pg = pfn_to_page((pa + count - 1) >> PAGE_SHIFT);
-	if (PageHighMem(pg)) {
-		p = ioremap_cache(pa, count);
-		if (!p)
-			return -ENXIO;
-	} else
-		p = __va(pa);
-
-	remain = copy_to_user(user_buf, p, count);
-
-	if (PageHighMem(pg))
-		iounmap(p);
-
-	if (remain)
-		return -EFAULT;
-
-	*ppos = pos + count;
-
-	return count;
-}
-
-static const struct file_operations fops_setup_data = {
-	.read		= setup_data_read,
-	.open		= simple_open,
-	.llseek		= default_llseek,
-};
-
-static int __init
-create_setup_data_node(struct dentry *parent, int no,
-		       struct setup_data_node *node)
-{
-	struct dentry *d, *type, *data;
-	char buf[16];
-
-	sprintf(buf, "%d", no);
-	d = debugfs_create_dir(buf, parent);
-	if (!d)
-		return -ENOMEM;
-
-	type = debugfs_create_x32("type", S_IRUGO, d, &node->type);
-	if (!type)
-		goto err_dir;
-
-	data = debugfs_create_file("data", S_IRUGO, d, node, &fops_setup_data);
-	if (!data)
-		goto err_type;
-
-	return 0;
-
-err_type:
-	debugfs_remove(type);
-err_dir:
-	debugfs_remove(d);
-	return -ENOMEM;
-}
-
-static int __init create_setup_data_nodes(struct dentry *parent)
-{
-	struct setup_data_node *node;
-	struct setup_data *data;
-	int error;
-	struct dentry *d;
-	struct page *pg;
-	u64 pa_data;
-	int no = 0;
-
-	d = debugfs_create_dir("setup_data", parent);
-	if (!d)
-		return -ENOMEM;
-
-	pa_data = boot_params.hdr.setup_data;
-
-	while (pa_data) {
-		node = kmalloc(sizeof(*node), GFP_KERNEL);
-		if (!node) {
-			error = -ENOMEM;
-			goto err_dir;
-		}
-
-		pg = pfn_to_page((pa_data+sizeof(*data)-1) >> PAGE_SHIFT);
-		if (PageHighMem(pg)) {
-			data = ioremap_cache(pa_data, sizeof(*data));
-			if (!data) {
-				kfree(node);
-				error = -ENXIO;
-				goto err_dir;
-			}
-		} else
-			data = __va(pa_data);
-
-		node->paddr = pa_data;
-		node->type = data->type;
-		node->len = data->len;
-		error = create_setup_data_node(d, no, node);
-		pa_data = data->next;
-
-		if (PageHighMem(pg))
-			iounmap(data);
-		if (error)
-			goto err_dir;
-		no++;
-	}
-
-	return 0;
-
-err_dir:
-	debugfs_remove(d);
-	return error;
-}
-
 static struct debugfs_blob_wrapper boot_params_blob = {
 	.data		= &boot_params,
 	.size		= sizeof(boot_params),
@@ -181,14 +45,8 @@ static int __init boot_params_kdebugfs_init(void)
 	if (!data)
 		goto err_version;
 
-	error = create_setup_data_nodes(dbp);
-	if (error)
-		goto err_data;
-
 	return 0;
 
-err_data:
-	debugfs_remove(data);
 err_version:
 	debugfs_remove(version);
 err_dir:
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index 94f95e0..a7e688c 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -486,20 +486,6 @@ static void __init parse_setup_data(void)
 	boot_params.hdr.setup_data = 0; /* all done */
 }
 
-static void __init memblock_x86_reserve_range_setup_data(void)
-{
-	struct setup_data *data;
-	u64 pa_data;
-
-	pa_data = boot_params.hdr.setup_data;
-	while (pa_data) {
-		data = early_memremap(pa_data, sizeof(*data));
-		memblock_reserve(pa_data, sizeof(*data) + data->len);
-		pa_data = data->next;
-		early_iounmap(data, sizeof(*data));
-	}
-}
-
 /*
  * --------- Crashkernel reservation ------------------------------
  */
@@ -1006,9 +992,6 @@ void __init setup_arch(char **cmdline_p)
 
 	x86_report_nx();
 
-	/* after early param, so could get panic from serial */
-	memblock_x86_reserve_range_setup_data();
-
 	if (acpi_mps_check()) {
 #ifdef CONFIG_X86_LOCAL_APIC
 		disable_apic = 1;
-- 
1.8.4.5

[PATCH v3 6/8] x86, boot, PCI: Convert SETUP_PCI data to list

[Yinghai Lu]

So we could avoid ioremap every time later.

Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: linux-pci@vger.kernel.org
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
---
 arch/x86/include/asm/pci.h |  2 ++
 arch/x86/kernel/setup.c    |  1 +
 arch/x86/pci/common.c      | 77 +++++++++++++++++++++++++++++++++++++---------
 3 files changed, 65 insertions(+), 15 deletions(-)

diff --git a/arch/x86/include/asm/pci.h b/arch/x86/include/asm/pci.h
index 7fbd5f3..99b261f 100644
--- a/arch/x86/include/asm/pci.h
+++ b/arch/x86/include/asm/pci.h
@@ -88,9 +88,11 @@ static inline void pci_dma_burst_advice(struct pci_dev *pdev,
 	*strategy_parameter = ~0UL;
 }
 void add_pci(u64 pa_data);
+int fill_setup_pci_entries(void);
 #else
 static inline void early_quirks(void) { }
 static inline void add_pci(u64 pa_data) { }
+static inline int fill_setup_pci_entries(void) { }
 #endif
 
 extern void pci_iommu_alloc(void);
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index a7e688c..39940a7 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -1195,6 +1195,7 @@ void __init setup_arch(char **cmdline_p)
 	acpi_boot_init();
 	sfi_init();
 	x86_dtb_init();
+	fill_setup_pci_entries();
 
 	/*
 	 * get boot-time SMP configuration:
diff --git a/arch/x86/pci/common.c b/arch/x86/pci/common.c
index f40df5c..67dc2a9 100644
--- a/arch/x86/pci/common.c
+++ b/arch/x86/pci/common.c
@@ -668,7 +668,7 @@ unsigned int pcibios_assign_all_busses(void)
 	return (pci_probe & PCI_ASSIGN_ALL_BUSSES) ? 1 : 0;
 }
 
-static u64 pci_setup_data;
+static u64 pci_setup_data __initdata;
 void __init add_pci(u64 pa_data)
 {
 	struct setup_data *data;
@@ -683,36 +683,83 @@ void __init add_pci(u64 pa_data)
 	early_memunmap(data, sizeof(*data));
 }
 
-int pcibios_add_device(struct pci_dev *dev)
+struct firmware_setup_pci_entry {
+	struct list_head list;
+	uint16_t vendor;
+	uint16_t devid;
+	uint64_t pcilen;
+	unsigned long segment;
+	unsigned long bus;
+	unsigned long device;
+	unsigned long function;
+	phys_addr_t romdata;
+};
+
+static LIST_HEAD(setup_pci_entries);
+
+int __init fill_setup_pci_entries(void)
 {
 	struct setup_data *data;
 	struct pci_setup_rom *rom;
+	struct firmware_setup_pci_entry *entry;
+	phys_addr_t pa_entry;
 	u64 pa_data;
 
 	pa_data = pci_setup_data;
 	while (pa_data) {
-		data = ioremap(pa_data, sizeof(*rom));
+		data  = early_memremap(pa_data, sizeof(*rom));
 		if (!data)
 			return -ENOMEM;
-
 		rom = (struct pci_setup_rom *)data;
 
-		if ((pci_domain_nr(dev->bus) == rom->segment) &&
-		    (dev->bus->number == rom->bus) &&
-		    (PCI_SLOT(dev->devfn) == rom->device) &&
-		    (PCI_FUNC(dev->devfn) == rom->function) &&
-		    (dev->vendor == rom->vendor) &&
-		    (dev->device == rom->devid)) {
-			dev->rom = pa_data +
-			      offsetof(struct pci_setup_rom, romdata);
-			dev->romlen = rom->pcilen;
+		pa_entry = memblock_alloc(sizeof(*entry), sizeof(long));
+		if (!pa_entry) {
+			early_memunmap(data, sizeof(*rom));
+			return -ENOMEM;
+		}
+
+		entry = phys_to_virt(pa_entry);
+		entry->segment = rom->segment;
+		entry->bus = rom->bus;
+		entry->device = rom->device;
+		entry->function = rom->function;
+		entry->vendor = rom->vendor;
+		entry->devid = rom->devid;
+		entry->pcilen = rom->pcilen;
+		entry->romdata = pa_data +
+				 offsetof(struct pci_setup_rom, romdata);
+
+		list_add(&entry->list, &setup_pci_entries);
+
+		memblock_free(pa_data, sizeof(*rom));
+		pa_data = data->next;
+		early_memunmap(data, sizeof(*rom));
+	}
+
+	pci_setup_data = 0;
+
+	return 0;
+}
+
+int pcibios_add_device(struct pci_dev *dev)
+{
+	struct firmware_setup_pci_entry *entry;
+
+	list_for_each_entry(entry, &setup_pci_entries, list) {
+		if ((pci_domain_nr(dev->bus) == entry->segment) &&
+		    (dev->bus->number == entry->bus) &&
+		    (PCI_SLOT(dev->devfn) == entry->device) &&
+		    (PCI_FUNC(dev->devfn) == entry->function) &&
+		    (dev->vendor == entry->vendor) &&
+		    (dev->device == entry->devid)) {
+			dev->rom = entry->romdata;
+			dev->romlen = entry->pcilen;
 			dev_printk(KERN_DEBUG, &dev->dev, "set rom to [%#010lx, %#010lx] via SETUP_PCI\n",
 				   (unsigned long)dev->rom,
 				   (unsigned long)(dev->rom + dev->romlen - 1));
 		}
-		pa_data = data->next;
-		iounmap(data);
 	}
+
 	return 0;
 }
 
-- 
1.8.4.5

[PATCH v3 7/8] x86, boot, PCI: Copy SETUP_PCI rom to kernel space

[Yinghai Lu]

As EFI stub code could put them high when on 32bit or with exactmap=
on 64bit conf.

Check if the range is mapped, otherwise allocate new one and have
the rom data copied. So we could access them directly.

Signed-off-by: Yinghai Lu <yinghai@kernel.org>
---
 arch/x86/pci/common.c | 47 +++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 45 insertions(+), 2 deletions(-)

diff --git a/arch/x86/pci/common.c b/arch/x86/pci/common.c
index 67dc2a9..15e1b3f 100644
--- a/arch/x86/pci/common.c
+++ b/arch/x86/pci/common.c
@@ -697,6 +697,48 @@ struct firmware_setup_pci_entry {
 
 static LIST_HEAD(setup_pci_entries);
 
+static phys_addr_t check_copy(phys_addr_t start, unsigned long size)
+{
+	unsigned long start_pfn = PFN_DOWN(start);
+	unsigned long end_pfn = PFN_UP(start + size);
+	unsigned char *p, *q;
+	phys_addr_t pa_p, pa_q;
+	long sz = size;
+
+	if (pfn_range_is_mapped(start_pfn, end_pfn))
+		return start;
+
+	/* allocate and copy */
+	pa_p = memblock_alloc(size, PAGE_SIZE);
+	if (!pa_p)
+		return start;
+
+	p = phys_to_virt(pa_p);
+
+	pa_q = start;
+	while (sz > 0) {
+		long chunk_size = 64<<10;
+
+		if (chunk_size > sz)
+			chunk_size = sz;
+
+		q = early_memremap(pa_q, chunk_size);
+		if (!q) {
+			memblock_free(pa_p, size);
+			return start;
+		}
+		memcpy(p, q, chunk_size);
+		early_memunmap(q, chunk_size);
+		p += chunk_size;
+		pa_q += chunk_size;
+		sz -= chunk_size;
+	}
+
+	memblock_free(start, size);
+
+	return pa_p;
+}
+
 int __init fill_setup_pci_entries(void)
 {
 	struct setup_data *data;
@@ -726,8 +768,9 @@ int __init fill_setup_pci_entries(void)
 		entry->vendor = rom->vendor;
 		entry->devid = rom->devid;
 		entry->pcilen = rom->pcilen;
-		entry->romdata = pa_data +
-				 offsetof(struct pci_setup_rom, romdata);
+		entry->romdata = check_copy(pa_data +
+				      offsetof(struct pci_setup_rom, romdata),
+				      rom->pcilen);
 
 		list_add(&entry->list, &setup_pci_entries);
 
-- 
1.8.4.5

[PATCH v3 8/8] x86, boot, PCI: Export SETUP_PCI data via sysfs

[Yinghai Lu]

So we could let kexec-tools to rebuild SETUP_PCI and pass it to
second kernel if needed.

Now kexec-tools already build SETUP_EFI and SETUP_E820EXT.

Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: linux-pci@vger.kernel.org
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
---
 arch/x86/pci/common.c | 175 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 175 insertions(+)

diff --git a/arch/x86/pci/common.c b/arch/x86/pci/common.c
index 15e1b3f..502e707 100644
--- a/arch/x86/pci/common.c
+++ b/arch/x86/pci/common.c
@@ -685,6 +685,8 @@ void __init add_pci(u64 pa_data)
 
 struct firmware_setup_pci_entry {
 	struct list_head list;
+	struct kobject kobj;
+	struct bin_attribute *rom_attr;
 	uint16_t vendor;
 	uint16_t devid;
 	uint64_t pcilen;
@@ -806,6 +808,179 @@ int pcibios_add_device(struct pci_dev *dev)
 	return 0;
 }
 
+#ifdef CONFIG_SYSFS
+static inline struct firmware_setup_pci_entry *
+to_setup_pci_entry(struct kobject *kobj)
+{
+	return container_of(kobj, struct firmware_setup_pci_entry, kobj);
+}
+
+static ssize_t vendor_show(struct firmware_setup_pci_entry *entry, char *buf)
+{
+	return snprintf(buf, PAGE_SIZE, "0x%04llx\n",
+			(unsigned long long)entry->vendor);
+}
+
+static ssize_t devid_show(struct firmware_setup_pci_entry *entry, char *buf)
+{
+	return snprintf(buf, PAGE_SIZE, "0x%04llx\n",
+			(unsigned long long)entry->devid);
+}
+
+static ssize_t pcilen_show(struct firmware_setup_pci_entry *entry, char *buf)
+{
+	return snprintf(buf, PAGE_SIZE, "0x%llx\n",
+			(unsigned long long)entry->pcilen);
+}
+
+static ssize_t segment_show(struct firmware_setup_pci_entry *entry, char *buf)
+{
+	return snprintf(buf, PAGE_SIZE, "0x%04llx\n",
+			(unsigned long long)entry->segment);
+}
+
+static ssize_t bus_show(struct firmware_setup_pci_entry *entry, char *buf)
+{
+	return snprintf(buf, PAGE_SIZE, "0x%02llx\n",
+			(unsigned long long)entry->bus);
+}
+
+static ssize_t device_show(struct firmware_setup_pci_entry *entry, char *buf)
+{
+	return snprintf(buf, PAGE_SIZE, "0x%02llx\n",
+			(unsigned long long)entry->device);
+}
+
+static ssize_t function_show(struct firmware_setup_pci_entry *entry, char *buf)
+{
+	return snprintf(buf, PAGE_SIZE, "0x%1llx\n",
+			(unsigned long long)entry->function);
+}
+
+struct setup_pci_attribute {
+	struct attribute attr;
+	ssize_t (*show)(struct firmware_setup_pci_entry *entry, char *buf);
+};
+
+static inline struct setup_pci_attribute *to_setup_pci_attr(
+							struct attribute *attr)
+{
+	return container_of(attr, struct setup_pci_attribute, attr);
+}
+
+static ssize_t setup_pci_attr_show(struct kobject *kobj,
+				   struct attribute *attr, char *buf)
+{
+	struct firmware_setup_pci_entry *entry = to_setup_pci_entry(kobj);
+	struct setup_pci_attribute *setup_pci_attr = to_setup_pci_attr(attr);
+
+	return setup_pci_attr->show(entry, buf);
+}
+
+static struct setup_pci_attribute setup_pci_vendor_attr = __ATTR_RO(vendor);
+static struct setup_pci_attribute setup_pci_devid_attr = __ATTR_RO(devid);
+static struct setup_pci_attribute setup_pci_pcilen_attr = __ATTR_RO(pcilen);
+static struct setup_pci_attribute setup_pci_segment_attr = __ATTR_RO(segment);
+static struct setup_pci_attribute setup_pci_bus_attr = __ATTR_RO(bus);
+static struct setup_pci_attribute setup_pci_device_attr = __ATTR_RO(device);
+static struct setup_pci_attribute setup_pci_function_attr = __ATTR_RO(function);
+
+/*
+ * These are default attributes that are added for every memmap entry.
+ */
+static struct attribute *def_attrs[] = {
+	&setup_pci_vendor_attr.attr,
+	&setup_pci_devid_attr.attr,
+	&setup_pci_pcilen_attr.attr,
+	&setup_pci_segment_attr.attr,
+	&setup_pci_bus_attr.attr,
+	&setup_pci_device_attr.attr,
+	&setup_pci_function_attr.attr,
+	NULL
+};
+
+static const struct sysfs_ops setup_pci_attr_ops = {
+	.show = setup_pci_attr_show,
+};
+
+static struct kobj_type __refdata setup_pci_ktype = {
+	.sysfs_ops      = &setup_pci_attr_ops,
+	.default_attrs  = def_attrs,
+};
+
+static ssize_t setup_pci_rom_read(struct file *filp, struct kobject *kobj,
+				  struct bin_attribute *bin_attr, char *buf,
+				  loff_t off, size_t count)
+{
+	struct firmware_setup_pci_entry *entry = to_setup_pci_entry(kobj);
+
+	if (off >= entry->pcilen)
+		count = 0;
+	else {
+		unsigned char *rom = phys_to_virt(entry->romdata);
+
+		if (off + count > entry->pcilen)
+			count = entry->pcilen - off;
+
+		memcpy(buf, rom + off, count);
+	}
+
+	return count;
+}
+
+static int __init add_sysfs_fw_setup_pci_entry(
+					struct firmware_setup_pci_entry *entry)
+{
+	int retval = 0;
+	static int setup_pci_entries_nr;
+	static struct kset *setup_pci_kset;
+	struct bin_attribute *attr;
+
+	kobject_init(&entry->kobj, &setup_pci_ktype);
+
+	if (!setup_pci_kset) {
+		setup_pci_kset = kset_create_and_add("setup_pci", NULL,
+						     firmware_kobj);
+		if (!setup_pci_kset)
+			return -ENOMEM;
+	}
+
+	entry->kobj.kset = setup_pci_kset;
+	retval = kobject_add(&entry->kobj, NULL, "%d", setup_pci_entries_nr++);
+	if (retval) {
+		kobject_put(&entry->kobj);
+		return retval;
+	}
+
+	attr = kzalloc(sizeof(*attr), GFP_ATOMIC);
+	if (!attr)
+		return -ENOMEM;
+
+	sysfs_bin_attr_init(attr);
+	attr->size = entry->pcilen;
+	attr->attr.name = "rom";
+	attr->attr.mode = S_IRUSR;
+	attr->read = setup_pci_rom_read;
+	retval = sysfs_create_bin_file(&entry->kobj, attr);
+	if (retval)
+		kfree(attr);
+	entry->rom_attr = attr;
+
+	return retval;
+}
+
+static int __init firmware_setup_pci_init(void)
+{
+	struct firmware_setup_pci_entry *entry;
+
+	list_for_each_entry(entry, &setup_pci_entries, list)
+		add_sysfs_fw_setup_pci_entry(entry);
+
+	return 0;
+}
+late_initcall(firmware_setup_pci_init);
+#endif
+
 int pcibios_enable_device(struct pci_dev *dev, int mask)
 {
 	int err;
-- 
1.8.4.5

Re: [PATCH v3 1/8] x86: Kill E820_RESERVED_KERN

[David Rientjes]

On Sat, 7 Mar 2015, Yinghai Lu wrote:

> Now we are using memblock to do early resource reserver/allocation
> instead of using e820 map directly, and setup_data is reserved in
> memblock early already.
> Also kexec generate setup_data and pass pointer to second kernel,
> so second kernel reserve setup_data by their own.
> (Now kexec-tools create SETUP_EFI and SETUP_E820_EXT).
> 
> We can kill E820_RESERVED_KERN and not touch e820 map at all.
> 
> That will fix bug in mark_nonsave_region that can not handle that
> case: E820_RAM and E820_RESERVED_KERN ranges are continuous and
> boundary is not page aligned.
> 
> Bugzilla: https://bugzilla.opensuse.org/show_bug.cgi?id=913885

Is this the bug referenced in the commit message that is fixed?  If so, 
it's only a bug for resume, correct?  I'm not sure if that's clear enough 
just from the commit message, I was looking at this patch for an e820 
problem I'm currently facing on 3.3.

> Reported-by: "Lee, Chun-Yi" <jlee@suse.com>
> Tested-by: "Lee, Chun-Yi" <jlee@suse.com>
> Cc: "Lee, Chun-Yi" <jlee@suse.com>
> Signed-off-by: Yinghai Lu <yinghai@kernel.org>
> Cc: stable@vger.kernel.org

Hmm, although the bug is reported for a 3.12 kernel, I assume this is for 
stable 3.10+?  If so, it should apply fine with the exception of removing 
e820_reserve_setup_data() from setup_arch() rather than 
memblock_x86_reserve_range_setup_data().  Or is it for 3.2 as well and 
needs to be completely rebased for that kernel?

Re: [PATCH v3 1/8] x86: Kill E820_RESERVED_KERN

[Yinghai Lu]

On Sat, Mar 7, 2015 at 5:59 PM, David Rientjes <rientjes@google.com> wrote:
>
> Hmm, although the bug is reported for a 3.12 kernel, I assume this is for
> stable 3.10+?  If so, it should apply fine with the exception of removing
> e820_reserve_setup_data() from setup_arch() rather than
> memblock_x86_reserve_range_setup_data().  Or is it for 3.2 as well and
> needs to be completely rebased for that kernel?

For 3.10+, we will need to following patches, otherwise will have warning for
SETUP_PCI with ioremap.

for 3.2 that does not SETUP_PCI, should be ok, but will need rebase.

Thanks

Yinghai

Re: [PATCH v3 1/8] x86: Kill E820_RESERVED_KERN

[joeyli]

On Sat, Mar 07, 2015 at 05:59:14PM -0800, David Rientjes wrote:
> On Sat, 7 Mar 2015, Yinghai Lu wrote:
> 
> > Now we are using memblock to do early resource reserver/allocation
> > instead of using e820 map directly, and setup_data is reserved in
> > memblock early already.
> > Also kexec generate setup_data and pass pointer to second kernel,
> > so second kernel reserve setup_data by their own.
> > (Now kexec-tools create SETUP_EFI and SETUP_E820_EXT).
> > 
> > We can kill E820_RESERVED_KERN and not touch e820 map at all.
> > 
> > That will fix bug in mark_nonsave_region that can not handle that
> > case: E820_RAM and E820_RESERVED_KERN ranges are continuous and
> > boundary is not page aligned.
> > 
> > Bugzilla: https://bugzilla.opensuse.org/show_bug.cgi?id=913885
> 
> Is this the bug referenced in the commit message that is fixed?  If so, 
> it's only a bug for resume, correct?  I'm not sure if that's clear enough 
> just from the commit message, I was looking at this patch for an e820 
> problem I'm currently facing on 3.3.

Yinghai's patches fixed the e820 not page aligned issue that's one of the
issues on bug reporter's machine. I found another issue of the BIOS that
sometimes it doesn't really keep the e820 table unchanging for hibernate
resuming, this BIOS issue causes the total available page number checking
fail. I will file another openSUSE bug to separate those 2 issues.

> 
> > Reported-by: "Lee, Chun-Yi" <jlee@suse.com>
> > Tested-by: "Lee, Chun-Yi" <jlee@suse.com>
> > Cc: "Lee, Chun-Yi" <jlee@suse.com>
> > Signed-off-by: Yinghai Lu <yinghai@kernel.org>
> > Cc: stable@vger.kernel.org
> 
> Hmm, although the bug is reported for a 3.12 kernel, I assume this is for 
> stable 3.10+?  If so, it should apply fine with the exception of removing 
> e820_reserve_setup_data() from setup_arch() rather than 
> memblock_x86_reserve_range_setup_data().  Or is it for 3.2 as well and 
> needs to be completely rebased for that kernel?

Thanks a lot!
Joey Lee