From 6428bc7bd3f35e43c8cb7359cb89d83248d339d2 Mon Sep 17 00:00:00 2001 From: Helge Deller Date: Wed, 30 Aug 2023 07:56:04 +0200 Subject: parisc: sba_iommu: Fix build warning if procfs if disabled Clean up the code, e.g. make proc_mckinley_root static, drop the now empty mckinley header file and remove some unneeded ifdefs around procfs functions. Signed-off-by: Helge Deller Reported-by: kernel test robot Closes: https://lore.kernel.org/oe-kbuild-all/202308300800.Jod4sHzM-lkp@intel.com/ Fixes: 77e0ddf097d6 ("parisc: ccio-dma: Create private runway procfs root entry") --- arch/parisc/include/asm/mckinley.h | 8 -------- 1 file changed, 8 deletions(-) delete mode 100644 arch/parisc/include/asm/mckinley.h (limited to 'arch') diff --git a/arch/parisc/include/asm/mckinley.h b/arch/parisc/include/asm/mckinley.h deleted file mode 100644 index 1314390b9034..000000000000 --- a/arch/parisc/include/asm/mckinley.h +++ /dev/null @@ -1,8 +0,0 @@ -/* SPDX-License-Identifier: GPL-2.0 */ -#ifndef ASM_PARISC_MCKINLEY_H -#define ASM_PARISC_MCKINLEY_H - -/* declared in arch/parisc/kernel/setup.c */ -extern struct proc_dir_entry * proc_mckinley_root; - -#endif /*ASM_PARISC_MCKINLEY_H*/ -- cgit v1.2.3 From eb3255ee8f6f4691471a28fbf22db5e8901116cd Mon Sep 17 00:00:00 2001 From: Helge Deller Date: Wed, 30 Aug 2023 08:10:01 +0200 Subject: parisc: sba: Fix compile warning wrt list of SBA devices Fix this makecheck warning: drivers/parisc/sba_iommu.c:98:19: warning: symbol 'sba_list' was not declared. Should it be static? Signed-off-by: Helge Deller --- arch/parisc/include/asm/ropes.h | 3 +++ 1 file changed, 3 insertions(+) (limited to 'arch') diff --git a/arch/parisc/include/asm/ropes.h b/arch/parisc/include/asm/ropes.h index fd96706c7234..d7b941cfbccd 100644 --- a/arch/parisc/include/asm/ropes.h +++ b/arch/parisc/include/asm/ropes.h @@ -86,6 +86,9 @@ struct sba_device { struct ioc ioc[MAX_IOC]; }; +/* list of SBA's in system, see drivers/parisc/sba_iommu.c */ +extern struct sba_device *sba_list; + #define ASTRO_RUNWAY_PORT 0x582 #define IKE_MERCED_PORT 0x803 #define REO_MERCED_PORT 0x804 -- cgit v1.2.3 From c1ebb94071cb4455177bafa619423acb3494d15d Mon Sep 17 00:00:00 2001 From: Helge Deller Date: Wed, 30 Aug 2023 11:49:57 +0200 Subject: parisc: sba-iommu: Fix sparse warnigs Fix sparse warnings, as pdir is __le64 *. Signed-off-by: Helge Deller --- arch/parisc/include/asm/ropes.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch') diff --git a/arch/parisc/include/asm/ropes.h b/arch/parisc/include/asm/ropes.h index d7b941cfbccd..e2d2d7e9bfde 100644 --- a/arch/parisc/include/asm/ropes.h +++ b/arch/parisc/include/asm/ropes.h @@ -29,7 +29,7 @@ struct ioc { void __iomem *ioc_hpa; /* I/O MMU base address */ char *res_map; /* resource map, bit == pdir entry */ - u64 *pdir_base; /* physical base address */ + __le64 *pdir_base; /* physical base address */ unsigned long ibase; /* pdir IOV Space base - shared w/lba_pci */ unsigned long imask; /* pdir IOV Space mask - shared w/lba_pci */ #ifdef ZX1_SUPPORT @@ -113,7 +113,7 @@ static inline int IS_PLUTO(struct parisc_device *d) { #define SBA_PDIR_VALID_BIT 0x8000000000000000ULL -#define SBA_AGPGART_COOKIE 0x0000badbadc0ffeeULL +#define SBA_AGPGART_COOKIE (__force __le64) 0x0000badbadc0ffeeULL #define SBA_FUNC_ID 0x0000 /* function id */ #define SBA_FCLASS 0x0008 /* function class, bist, header, rev... */ -- cgit v1.2.3 From b137b9d60b8add5620a06c687a71ce18776730b0 Mon Sep 17 00:00:00 2001 From: Helge Deller Date: Thu, 31 Aug 2023 22:08:32 +0200 Subject: parisc: drivers: Fix sparse warning Fix "warning: directive in macro's argument list" warning. Signed-off-by: Helge Deller --- arch/parisc/kernel/drivers.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch') diff --git a/arch/parisc/kernel/drivers.c b/arch/parisc/kernel/drivers.c index 8f4b77648491..ed8b75948061 100644 --- a/arch/parisc/kernel/drivers.c +++ b/arch/parisc/kernel/drivers.c @@ -925,9 +925,9 @@ static __init void qemu_header(void) pr_info("#define PARISC_MODEL \"%s\"\n\n", boot_cpu_data.pdc.sys_model_name); + #define p ((unsigned long *)&boot_cpu_data.pdc.model) pr_info("#define PARISC_PDC_MODEL 0x%lx, 0x%lx, 0x%lx, " "0x%lx, 0x%lx, 0x%lx, 0x%lx, 0x%lx, 0x%lx\n\n", - #define p ((unsigned long *)&boot_cpu_data.pdc.model) p[0], p[1], p[2], p[3], p[4], p[5], p[6], p[7], p[8]); #undef p -- cgit v1.2.3 From b1bef1388c427cdad7331a9c8eb4ebbbe5b954b0 Mon Sep 17 00:00:00 2001 From: Helge Deller Date: Thu, 31 Aug 2023 22:36:12 +0200 Subject: parisc: irq: Make irq_stack_union static to avoid sparse warning Signed-off-by: Helge Deller --- arch/parisc/kernel/irq.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch') diff --git a/arch/parisc/kernel/irq.c b/arch/parisc/kernel/irq.c index 12c4d4104ade..2f81bfd4f15e 100644 --- a/arch/parisc/kernel/irq.c +++ b/arch/parisc/kernel/irq.c @@ -365,7 +365,7 @@ union irq_stack_union { volatile unsigned int lock[1]; }; -DEFINE_PER_CPU(union irq_stack_union, irq_stack_union) = { +static DEFINE_PER_CPU(union irq_stack_union, irq_stack_union) = { .slock = { 1,1,1,1 }, }; #endif -- cgit v1.2.3 From 3f091387a39795812aab4303949bbc9baa22c077 Mon Sep 17 00:00:00 2001 From: Helge Deller Date: Fri, 1 Sep 2023 08:04:31 +0200 Subject: parisc: shmparam.h: Document aliasing requirements of PA-RISC Add some documentation why PA-RISC uses SHMLBA and SHM_COLOUR. Signed-off-by: Helge Deller --- arch/parisc/include/asm/shmparam.h | 15 +++++++++++++++ 1 file changed, 15 insertions(+) (limited to 'arch') diff --git a/arch/parisc/include/asm/shmparam.h b/arch/parisc/include/asm/shmparam.h index 74f74e4d35b7..5a95b0f62b87 100644 --- a/arch/parisc/include/asm/shmparam.h +++ b/arch/parisc/include/asm/shmparam.h @@ -2,6 +2,21 @@ #ifndef _ASMPARISC_SHMPARAM_H #define _ASMPARISC_SHMPARAM_H +/* + * PA-RISC uses virtually indexed & physically tagged (VIPT) caches + * which has strict requirements when two pages to the same physical + * address are accessed through different mappings. Read the section + * "Address Aliasing" in the arch docs for more detail: + * PA-RISC 1.1 (page 3-6): + * https://parisc.wiki.kernel.org/images-parisc/6/68/Pa11_acd.pdf + * PA-RISC 2.0 (page F-5): + * https://parisc.wiki.kernel.org/images-parisc/7/73/Parisc2.0.pdf + * + * For Linux we allow kernel and userspace to map pages on page size + * granularity (SHMLBA) but have to ensure that, if two pages are + * mapped to the same physical address, the virtual and physical + * addresses modulo SHM_COLOUR are identical. + */ #define SHMLBA PAGE_SIZE /* attach addr a multiple of this */ #define SHM_COLOUR 0x00400000 /* shared mappings colouring */ -- cgit v1.2.3 From 70bd68d5b61a63f87b5e986d9100c8ce46c5df4d Mon Sep 17 00:00:00 2001 From: Helge Deller Date: Fri, 1 Sep 2023 16:09:23 +0200 Subject: parisc: Prepare for Block-TLB support on 32-bit kernel Change HUGEPAGE_SIZE to become 4 MB on 32-bit kernels, which leads that kernel code and kernel data will start on 4 MB boundaries. Although a 32-bit kernel does not support huge pages, most machines have support for Block-TLBs (BTLB) which allow to configure the system to use large pages (block TLBs) to minimize the TLB contention. This is done through calls to PDC and the 32-bit kernel can then call BTLB PDC functions to tell the machine to optimize the TLBs. Signed-off-by: Helge Deller --- arch/parisc/kernel/asm-offsets.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch') diff --git a/arch/parisc/kernel/asm-offsets.c b/arch/parisc/kernel/asm-offsets.c index 94652e13c260..757816a7bd4b 100644 --- a/arch/parisc/kernel/asm-offsets.c +++ b/arch/parisc/kernel/asm-offsets.c @@ -275,6 +275,8 @@ int main(void) * and kernel data on physical huge pages */ #ifdef CONFIG_HUGETLB_PAGE DEFINE(HUGEPAGE_SIZE, 1UL << REAL_HPAGE_SHIFT); +#elif !defined(CONFIG_64BIT) + DEFINE(HUGEPAGE_SIZE, 4*1024*1024); #else DEFINE(HUGEPAGE_SIZE, PAGE_SIZE); #endif -- cgit v1.2.3 From eda205211a522312b667d5bd25d58bee8504c09e Mon Sep 17 00:00:00 2001 From: Helge Deller Date: Thu, 7 Sep 2023 08:03:35 +0200 Subject: parisc: BTLB: Clear possibly existing BTLB entries Call PDC to remove all existing BTLB entries (which may exist from some previous operating system runs) before switching to virtual mode. Signed-off-by: Helge Deller --- arch/parisc/kernel/head.S | 16 +++++++++++++--- 1 file changed, 13 insertions(+), 3 deletions(-) (limited to 'arch') diff --git a/arch/parisc/kernel/head.S b/arch/parisc/kernel/head.S index fd15fd4bbb61..a171bf3c6b31 100644 --- a/arch/parisc/kernel/head.S +++ b/arch/parisc/kernel/head.S @@ -180,10 +180,10 @@ $pgt_fill_loop: std %dp,0x18(%r10) #endif -#ifdef CONFIG_64BIT - /* Get PDCE_PROC for monarch CPU. */ #define MEM_PDC_LO 0x388 #define MEM_PDC_HI 0x35C +#ifdef CONFIG_64BIT + /* Get PDCE_PROC for monarch CPU. */ ldw MEM_PDC_LO(%r0),%r3 ldw MEM_PDC_HI(%r0),%r10 depd %r10, 31, 32, %r3 /* move to upper word */ @@ -269,7 +269,17 @@ stext_pdc_ret: tovirt_r1 %r6 mtctl %r6,%cr30 /* restore task thread info */ #endif - + +#ifndef CONFIG_64BIT + /* clear all BTLBs */ + ldi PDC_BLOCK_TLB,%arg0 + load32 PA(stext_pdc_btlb_ret), %rp + ldw MEM_PDC_LO(%r0),%r3 + bv (%r3) + ldi PDC_BTLB_PURGE_ALL,%arg1 +stext_pdc_btlb_ret: +#endif + /* PARANOID: clear user scratch/user space SR's */ mtsp %r0,%sr0 mtsp %r0,%sr1 -- cgit v1.2.3 From 510610f96d65277940a02f47d7bc7a06c8a2ab7a Mon Sep 17 00:00:00 2001 From: Helge Deller Date: Thu, 7 Sep 2023 08:06:37 +0200 Subject: parisc: BTLB: Add BTLB insert and purge firmware function wrappers Signed-off-by: Helge Deller --- arch/parisc/include/asm/pdc.h | 3 +++ arch/parisc/kernel/firmware.c | 42 +++++++++++++++++++++++++++++++++--------- 2 files changed, 36 insertions(+), 9 deletions(-) (limited to 'arch') diff --git a/arch/parisc/include/asm/pdc.h b/arch/parisc/include/asm/pdc.h index 269b9a159f01..2fd7eb2f3746 100644 --- a/arch/parisc/include/asm/pdc.h +++ b/arch/parisc/include/asm/pdc.h @@ -46,6 +46,9 @@ int pdc_cache_info(struct pdc_cache_info *cache); int pdc_spaceid_bits(unsigned long *space_bits); #ifndef CONFIG_PA20 int pdc_btlb_info(struct pdc_btlb_info *btlb); +int pdc_btlb_insert(unsigned long long vpage, unsigned long physpage, unsigned long len, + unsigned long entry_info, unsigned long slot); +int pdc_btlb_purge_all(void); int pdc_mem_map_hpa(struct pdc_memory_map *r_addr, struct pdc_module_path *mod_path); #endif /* !CONFIG_PA20 */ int pdc_pim_toc11(struct pdc_toc_pim_11 *ret); diff --git a/arch/parisc/kernel/firmware.c b/arch/parisc/kernel/firmware.c index 8f37e75f2fb9..79bf44d04aa8 100644 --- a/arch/parisc/kernel/firmware.c +++ b/arch/parisc/kernel/firmware.c @@ -696,18 +696,42 @@ int pdc_spaceid_bits(unsigned long *space_bits) */ int pdc_btlb_info(struct pdc_btlb_info *btlb) { - int retval; + int retval; unsigned long flags; - spin_lock_irqsave(&pdc_lock, flags); - retval = mem_pdc_call(PDC_BLOCK_TLB, PDC_BTLB_INFO, __pa(pdc_result), 0); - memcpy(btlb, pdc_result, sizeof(*btlb)); - spin_unlock_irqrestore(&pdc_lock, flags); + spin_lock_irqsave(&pdc_lock, flags); + retval = mem_pdc_call(PDC_BLOCK_TLB, PDC_BTLB_INFO, __pa(pdc_result), 0); + memcpy(btlb, pdc_result, sizeof(*btlb)); + spin_unlock_irqrestore(&pdc_lock, flags); - if(retval < 0) { - btlb->max_size = 0; - } - return retval; + if(retval < 0) { + btlb->max_size = 0; + } + return retval; +} + +int pdc_btlb_insert(unsigned long long vpage, unsigned long physpage, unsigned long len, + unsigned long entry_info, unsigned long slot) +{ + int retval; + unsigned long flags; + + spin_lock_irqsave(&pdc_lock, flags); + retval = mem_pdc_call(PDC_BLOCK_TLB, PDC_BTLB_INSERT, (unsigned long) (vpage >> 32), + (unsigned long) vpage, physpage, len, entry_info, slot); + spin_unlock_irqrestore(&pdc_lock, flags); + return retval; +} + +int pdc_btlb_purge_all(void) +{ + int retval; + unsigned long flags; + + spin_lock_irqsave(&pdc_lock, flags); + retval = mem_pdc_call(PDC_BLOCK_TLB, PDC_BTLB_PURGE_ALL); + spin_unlock_irqrestore(&pdc_lock, flags); + return retval; } /** -- cgit v1.2.3 From 4695e45ec0cd4d422694b69e9f39c742bb1f35fc Mon Sep 17 00:00:00 2001 From: Helge Deller Date: Thu, 7 Sep 2023 08:07:55 +0200 Subject: parisc: BTLB: _edata symbol has to be page aligned for BTLB support Signed-off-by: Helge Deller --- arch/parisc/kernel/vmlinux.lds.S | 1 + 1 file changed, 1 insertion(+) (limited to 'arch') diff --git a/arch/parisc/kernel/vmlinux.lds.S b/arch/parisc/kernel/vmlinux.lds.S index 1aaa2ca09800..58694d1989c2 100644 --- a/arch/parisc/kernel/vmlinux.lds.S +++ b/arch/parisc/kernel/vmlinux.lds.S @@ -154,6 +154,7 @@ SECTIONS } /* End of data section */ + . = ALIGN(PAGE_SIZE); _edata = .; /* BSS */ -- cgit v1.2.3 From 3756597a684da9fbf966f91ed351033ac1a0f55f Mon Sep 17 00:00:00 2001 From: Helge Deller Date: Thu, 7 Sep 2023 08:51:20 +0200 Subject: parisc: firmware: Simplify calling non-PA20 functions Instead of usig #ifdefs, simply return PDC_BAD_PROC for functions which aren't available on 64-bit CPUs. Signed-off-by: Helge Deller --- arch/parisc/include/asm/pdc.h | 2 -- arch/parisc/kernel/firmware.c | 14 ++++++++++++-- 2 files changed, 12 insertions(+), 4 deletions(-) (limited to 'arch') diff --git a/arch/parisc/include/asm/pdc.h b/arch/parisc/include/asm/pdc.h index 2fd7eb2f3746..5d2d9737e579 100644 --- a/arch/parisc/include/asm/pdc.h +++ b/arch/parisc/include/asm/pdc.h @@ -44,13 +44,11 @@ int pdc_model_capabilities(unsigned long *capabilities); int pdc_model_platform_info(char *orig_prod_num, char *current_prod_num, char *serial_no); int pdc_cache_info(struct pdc_cache_info *cache); int pdc_spaceid_bits(unsigned long *space_bits); -#ifndef CONFIG_PA20 int pdc_btlb_info(struct pdc_btlb_info *btlb); int pdc_btlb_insert(unsigned long long vpage, unsigned long physpage, unsigned long len, unsigned long entry_info, unsigned long slot); int pdc_btlb_purge_all(void); int pdc_mem_map_hpa(struct pdc_memory_map *r_addr, struct pdc_module_path *mod_path); -#endif /* !CONFIG_PA20 */ int pdc_pim_toc11(struct pdc_toc_pim_11 *ret); int pdc_pim_toc20(struct pdc_toc_pim_20 *ret); int pdc_lan_station_id(char *lan_addr, unsigned long net_hpa); diff --git a/arch/parisc/kernel/firmware.c b/arch/parisc/kernel/firmware.c index 79bf44d04aa8..81078abec521 100644 --- a/arch/parisc/kernel/firmware.c +++ b/arch/parisc/kernel/firmware.c @@ -687,7 +687,6 @@ int pdc_spaceid_bits(unsigned long *space_bits) return retval; } -#ifndef CONFIG_PA20 /** * pdc_btlb_info - Return block TLB information. * @btlb: The return buffer. @@ -699,6 +698,9 @@ int pdc_btlb_info(struct pdc_btlb_info *btlb) int retval; unsigned long flags; + if (IS_ENABLED(CONFIG_PA20)) + return PDC_BAD_PROC; + spin_lock_irqsave(&pdc_lock, flags); retval = mem_pdc_call(PDC_BLOCK_TLB, PDC_BTLB_INFO, __pa(pdc_result), 0); memcpy(btlb, pdc_result, sizeof(*btlb)); @@ -716,6 +718,9 @@ int pdc_btlb_insert(unsigned long long vpage, unsigned long physpage, unsigned l int retval; unsigned long flags; + if (IS_ENABLED(CONFIG_PA20)) + return PDC_BAD_PROC; + spin_lock_irqsave(&pdc_lock, flags); retval = mem_pdc_call(PDC_BLOCK_TLB, PDC_BTLB_INSERT, (unsigned long) (vpage >> 32), (unsigned long) vpage, physpage, len, entry_info, slot); @@ -728,6 +733,9 @@ int pdc_btlb_purge_all(void) int retval; unsigned long flags; + if (IS_ENABLED(CONFIG_PA20)) + return PDC_BAD_PROC; + spin_lock_irqsave(&pdc_lock, flags); retval = mem_pdc_call(PDC_BLOCK_TLB, PDC_BTLB_PURGE_ALL); spin_unlock_irqrestore(&pdc_lock, flags); @@ -752,6 +760,9 @@ int pdc_mem_map_hpa(struct pdc_memory_map *address, int retval; unsigned long flags; + if (IS_ENABLED(CONFIG_PA20)) + return PDC_BAD_PROC; + spin_lock_irqsave(&pdc_lock, flags); memcpy(pdc_result2, mod_path, sizeof(*mod_path)); retval = mem_pdc_call(PDC_MEM_MAP, PDC_MEM_MAP_HPA, __pa(pdc_result), @@ -761,7 +772,6 @@ int pdc_mem_map_hpa(struct pdc_memory_map *address, return retval; } -#endif /* !CONFIG_PA20 */ /** * pdc_lan_station_id - Get the LAN address. -- cgit v1.2.3 From e5ef93d02d6c9cc3a14e7348481c9e41a528caa1 Mon Sep 17 00:00:00 2001 From: Helge Deller Date: Thu, 7 Sep 2023 08:57:44 +0200 Subject: parisc: BTLB: Initialize BTLB tables at CPU startup Initialize the BTLB entries when starting up a CPU. Note that BTLBs are not available on 64-bit CPUs. Signed-off-by: Helge Deller --- arch/parisc/include/asm/cache.h | 1 + arch/parisc/include/asm/processor.h | 1 + arch/parisc/kernel/cache.c | 8 +---- arch/parisc/kernel/processor.c | 2 ++ arch/parisc/mm/init.c | 72 +++++++++++++++++++++++++++++++++++++ 5 files changed, 77 insertions(+), 7 deletions(-) (limited to 'arch') diff --git a/arch/parisc/include/asm/cache.h b/arch/parisc/include/asm/cache.h index e23d06b51a20..2a60d7a72f1f 100644 --- a/arch/parisc/include/asm/cache.h +++ b/arch/parisc/include/asm/cache.h @@ -37,6 +37,7 @@ extern int split_tlb; extern int dcache_stride; extern int icache_stride; extern struct pdc_cache_info cache_info; +extern struct pdc_btlb_info btlb_info; void parisc_setup_cache_timing(void); #define pdtlb(sr, addr) asm volatile("pdtlb 0(%%sr%0,%1)" \ diff --git a/arch/parisc/include/asm/processor.h b/arch/parisc/include/asm/processor.h index d77c43d32974..ff6cbdb6903b 100644 --- a/arch/parisc/include/asm/processor.h +++ b/arch/parisc/include/asm/processor.h @@ -310,6 +310,7 @@ extern void do_syscall_trace_exit(struct pt_regs *); struct seq_file; extern void early_trap_init(void); extern void collect_boot_cpu_data(void); +extern void btlb_init_per_cpu(void); extern int show_cpuinfo (struct seq_file *m, void *v); /* driver code in driver/parisc */ diff --git a/arch/parisc/kernel/cache.c b/arch/parisc/kernel/cache.c index 442109a48940..268d90a9325b 100644 --- a/arch/parisc/kernel/cache.c +++ b/arch/parisc/kernel/cache.c @@ -58,7 +58,7 @@ int pa_serialize_tlb_flushes __ro_after_init; struct pdc_cache_info cache_info __ro_after_init; #ifndef CONFIG_PA20 -static struct pdc_btlb_info btlb_info __ro_after_init; +struct pdc_btlb_info btlb_info __ro_after_init; #endif DEFINE_STATIC_KEY_TRUE(parisc_has_cache); @@ -264,12 +264,6 @@ parisc_cache_init(void) icache_stride = CAFL_STRIDE(cache_info.ic_conf); #undef CAFL_STRIDE -#ifndef CONFIG_PA20 - if (pdc_btlb_info(&btlb_info) < 0) { - memset(&btlb_info, 0, sizeof btlb_info); - } -#endif - if ((boot_cpu_data.pdc.capabilities & PDC_MODEL_NVA_MASK) == PDC_MODEL_NVA_UNSUPPORTED) { printk(KERN_WARNING "parisc_cache_init: Only equivalent aliasing supported!\n"); diff --git a/arch/parisc/kernel/processor.c b/arch/parisc/kernel/processor.c index a0e2d37c5b3b..1fc89fa2c2d2 100644 --- a/arch/parisc/kernel/processor.c +++ b/arch/parisc/kernel/processor.c @@ -368,6 +368,8 @@ int init_per_cpu(int cpunum) /* FUTURE: Enable Performance Monitor : ccr bit 0x20 */ init_percpu_prof(cpunum); + btlb_init_per_cpu(); + return ret; } diff --git a/arch/parisc/mm/init.c b/arch/parisc/mm/init.c index a088c243edea..a2a3e89f2d9a 100644 --- a/arch/parisc/mm/init.c +++ b/arch/parisc/mm/init.c @@ -32,6 +32,7 @@ #include #include #include +#include extern int data_start; extern void parisc_kernel_start(void); /* Kernel entry point in head.S */ @@ -720,6 +721,77 @@ void __init paging_init(void) parisc_bootmem_free(); } +static void alloc_btlb(unsigned long start, unsigned long end, int *slot, + unsigned long entry_info) +{ + const int slot_max = btlb_info.fixed_range_info.num_comb; + int min_num_pages = btlb_info.min_size; + unsigned long size; + + /* map at minimum 4 pages */ + if (min_num_pages < 4) + min_num_pages = 4; + + size = HUGEPAGE_SIZE; + while (start < end && *slot < slot_max && size >= PAGE_SIZE) { + /* starting address must have same alignment as size! */ + /* if correctly aligned and fits in double size, increase */ + if (((start & (2 * size - 1)) == 0) && + (end - start) >= (2 * size)) { + size <<= 1; + continue; + } + /* if current size alignment is too big, try smaller size */ + if ((start & (size - 1)) != 0) { + size >>= 1; + continue; + } + if ((end - start) >= size) { + if ((size >> PAGE_SHIFT) >= min_num_pages) + pdc_btlb_insert(start >> PAGE_SHIFT, __pa(start) >> PAGE_SHIFT, + size >> PAGE_SHIFT, entry_info, *slot); + (*slot)++; + start += size; + continue; + } + size /= 2; + continue; + } +} + +void btlb_init_per_cpu(void) +{ + unsigned long s, t, e; + int slot; + + /* BTLBs are not available on 64-bit CPUs */ + if (IS_ENABLED(CONFIG_PA20)) + return; + else if (pdc_btlb_info(&btlb_info) < 0) { + memset(&btlb_info, 0, sizeof btlb_info); + } + + /* insert BLTLBs for code and data segments */ + s = (uintptr_t) dereference_function_descriptor(&_stext); + e = (uintptr_t) dereference_function_descriptor(&_etext); + t = (uintptr_t) dereference_function_descriptor(&_sdata); + BUG_ON(t != e); + + /* code segments */ + slot = 0; + alloc_btlb(s, e, &slot, 0x13800000); + + /* sanity check */ + t = (uintptr_t) dereference_function_descriptor(&_edata); + e = (uintptr_t) dereference_function_descriptor(&__bss_start); + BUG_ON(t != e); + + /* data segments */ + s = (uintptr_t) dereference_function_descriptor(&_sdata); + e = (uintptr_t) dereference_function_descriptor(&__bss_stop); + alloc_btlb(s, e, &slot, 0x11800000); +} + #ifdef CONFIG_PA20 /* -- cgit v1.2.3 From 762f169f5d9b92f057ad2c27ec4e3849b743239a Mon Sep 17 00:00:00 2001 From: Ard Biesheuvel Date: Mon, 7 Aug 2023 18:21:20 +0200 Subject: efi/x86: Move EFI runtime call setup/teardown helpers out of line Only the arch_efi_call_virt() macro that some architectures override needs to be a macro, given that it is variadic and encapsulates calls via function pointers that have different prototypes. The associated setup and teardown code are not special in this regard, and don't need to be instantiated at each call site. So turn them into ordinary C functions and move them out of line. Signed-off-by: Ard Biesheuvel --- arch/x86/include/asm/efi.h | 32 ++------------------------------ arch/x86/platform/efi/efi_32.c | 12 ++++++++++++ arch/x86/platform/efi/efi_64.c | 19 +++++++++++++++++-- 3 files changed, 31 insertions(+), 32 deletions(-) (limited to 'arch') diff --git a/arch/x86/include/asm/efi.h b/arch/x86/include/asm/efi.h index b0994ae3bc23..c4555b269a1b 100644 --- a/arch/x86/include/asm/efi.h +++ b/arch/x86/include/asm/efi.h @@ -91,19 +91,6 @@ static inline void efi_fpu_end(void) #ifdef CONFIG_X86_32 #define EFI_X86_KERNEL_ALLOC_LIMIT (SZ_512M - 1) - -#define arch_efi_call_virt_setup() \ -({ \ - efi_fpu_begin(); \ - firmware_restrict_branch_speculation_start(); \ -}) - -#define arch_efi_call_virt_teardown() \ -({ \ - firmware_restrict_branch_speculation_end(); \ - efi_fpu_end(); \ -}) - #else /* !CONFIG_X86_32 */ #define EFI_X86_KERNEL_ALLOC_LIMIT EFI_ALLOC_LIMIT @@ -116,14 +103,6 @@ extern bool efi_disable_ibt_for_runtime; __efi_call(__VA_ARGS__); \ }) -#define arch_efi_call_virt_setup() \ -({ \ - efi_sync_low_kernel_mappings(); \ - efi_fpu_begin(); \ - firmware_restrict_branch_speculation_start(); \ - efi_enter_mm(); \ -}) - #undef arch_efi_call_virt #define arch_efi_call_virt(p, f, args...) ({ \ u64 ret, ibt = ibt_save(efi_disable_ibt_for_runtime); \ @@ -132,13 +111,6 @@ extern bool efi_disable_ibt_for_runtime; ret; \ }) -#define arch_efi_call_virt_teardown() \ -({ \ - efi_leave_mm(); \ - firmware_restrict_branch_speculation_end(); \ - efi_fpu_end(); \ -}) - #ifdef CONFIG_KASAN /* * CONFIG_KASAN may redefine memset to __memset. __memset function is present @@ -168,8 +140,8 @@ extern void efi_delete_dummy_variable(void); extern void efi_crash_gracefully_on_page_fault(unsigned long phys_addr); extern void efi_free_boot_services(void); -void efi_enter_mm(void); -void efi_leave_mm(void); +void arch_efi_call_virt_setup(void); +void arch_efi_call_virt_teardown(void); /* kexec external ABI */ struct efi_setup_data { diff --git a/arch/x86/platform/efi/efi_32.c b/arch/x86/platform/efi/efi_32.c index e06a199423c0..b2cc7b4552a1 100644 --- a/arch/x86/platform/efi/efi_32.c +++ b/arch/x86/platform/efi/efi_32.c @@ -140,3 +140,15 @@ void __init efi_runtime_update_mappings(void) } } } + +void arch_efi_call_virt_setup(void) +{ + efi_fpu_begin(); + firmware_restrict_branch_speculation_start(); +} + +void arch_efi_call_virt_teardown(void) +{ + firmware_restrict_branch_speculation_end(); + efi_fpu_end(); +} diff --git a/arch/x86/platform/efi/efi_64.c b/arch/x86/platform/efi/efi_64.c index 77f7ac3668cb..91d31ac422d6 100644 --- a/arch/x86/platform/efi/efi_64.c +++ b/arch/x86/platform/efi/efi_64.c @@ -474,19 +474,34 @@ void __init efi_dump_pagetable(void) * can not change under us. * It should be ensured that there are no concurrent calls to this function. */ -void efi_enter_mm(void) +static void efi_enter_mm(void) { efi_prev_mm = current->active_mm; current->active_mm = &efi_mm; switch_mm(efi_prev_mm, &efi_mm, NULL); } -void efi_leave_mm(void) +static void efi_leave_mm(void) { current->active_mm = efi_prev_mm; switch_mm(&efi_mm, efi_prev_mm, NULL); } +void arch_efi_call_virt_setup(void) +{ + efi_sync_low_kernel_mappings(); + efi_fpu_begin(); + firmware_restrict_branch_speculation_start(); + efi_enter_mm(); +} + +void arch_efi_call_virt_teardown(void) +{ + efi_leave_mm(); + firmware_restrict_branch_speculation_end(); + efi_fpu_end(); +} + static DEFINE_SPINLOCK(efi_runtime_lock); /* -- cgit v1.2.3 From aba7e066c738d4b349413a271b2a236aa55bacbc Mon Sep 17 00:00:00 2001 From: Ard Biesheuvel Date: Wed, 2 Aug 2023 17:17:04 +0200 Subject: efi/x86: Ensure that EFI_RUNTIME_MAP is enabled for kexec CONFIG_EFI_RUNTIME_MAP needs to be enabled in order for kexec to be able to provide the required information about the EFI runtime mappings to the incoming kernel, regardless of whether kexec_load() or kexec_file_load() is being used. Without this information, kexec boot in EFI mode is not possible. The CONFIG_EFI_RUNTIME_MAP option is currently directly configurable if CONFIG_EXPERT is enabled, so that it can be turned on for debugging purposes even if KEXEC is not enabled. However, the upshot of this is that it can also be disabled even when it shouldn't. So tweak the Kconfig declarations to avoid this situation. Reported-by: Kirill A. Shutemov Signed-off-by: Ard Biesheuvel --- arch/x86/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index 982b777eadc7..66bfabae8814 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -1945,6 +1945,7 @@ config EFI select UCS2_STRING select EFI_RUNTIME_WRAPPERS select ARCH_USE_MEMREMAP_PROT + select EFI_RUNTIME_MAP if KEXEC_CORE help This enables the kernel to use EFI runtime services that are available (such as the EFI variable services). @@ -2020,7 +2021,6 @@ config EFI_MAX_FAKE_MEM config EFI_RUNTIME_MAP bool "Export EFI runtime maps to sysfs" if EXPERT depends on EFI - default KEXEC_CORE help Export EFI runtime memory regions to /sys/firmware/efi/runtime-map. That memory map is required by the 2nd kernel to set up EFI virtual -- cgit v1.2.3 From 5290e88ba2c742ca77c5f5b690e5af549cfd8591 Mon Sep 17 00:00:00 2001 From: Steve Wahl Date: Mon, 7 Aug 2023 09:17:30 -0500 Subject: x86/platform/uv: Use alternate source for socket to node data The UV code attempts to build a set of tables to allow it to do bidirectional socket<=>node lookups. But when nr_cpus is set to a smaller number than actually present, the cpu_to_node() mapping information for unused CPUs is not available to build_socket_tables(). This results in skipping some nodes or sockets when creating the tables and leaving some -1's for later code to trip. over, causing oopses. The problem is that the socket<=>node lookups are created by doing a loop over all CPUs, then looking up the CPU's APICID and socket. But if a CPU is not present, there is no way to start this lookup. Instead of looping over all CPUs, take CPUs out of the equation entirely. Loop over all APICIDs which are mapped to a valid NUMA node. Then just extract the socket-id from the APICID. This avoid tripping over disabled CPUs. Fixes: 8a50c5851927 ("x86/platform/uv: UV support for sub-NUMA clustering") Signed-off-by: Steve Wahl Signed-off-by: Dave Hansen Cc: stable@vger.kernel.org Link: https://lore.kernel.org/all/20230807141730.1117278-1-steve.wahl%40hpe.com --- arch/x86/kernel/apic/x2apic_uv_x.c | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-) (limited to 'arch') diff --git a/arch/x86/kernel/apic/x2apic_uv_x.c b/arch/x86/kernel/apic/x2apic_uv_x.c index d9f5d7492f83..205cee567629 100644 --- a/arch/x86/kernel/apic/x2apic_uv_x.c +++ b/arch/x86/kernel/apic/x2apic_uv_x.c @@ -1533,7 +1533,7 @@ static void __init build_socket_tables(void) { struct uv_gam_range_entry *gre = uv_gre_table; int nums, numn, nump; - int cpu, i, lnid; + int i, lnid, apicid; int minsock = _min_socket; int maxsock = _max_socket; int minpnode = _min_pnode; @@ -1584,15 +1584,14 @@ static void __init build_socket_tables(void) /* Set socket -> node values: */ lnid = NUMA_NO_NODE; - for_each_possible_cpu(cpu) { - int nid = cpu_to_node(cpu); - int apicid, sockid; + for (apicid = 0; apicid < ARRAY_SIZE(__apicid_to_node); apicid++) { + int nid = __apicid_to_node[apicid]; + int sockid; - if (lnid == nid) + if ((nid == NUMA_NO_NODE) || (lnid == nid)) continue; lnid = nid; - apicid = per_cpu(x86_cpu_to_apicid, cpu); sockid = apicid >> uv_cpuid.socketid_shift; if (_socket_to_node[sockid - minsock] == SOCK_EMPTY) -- cgit v1.2.3 From 25e73b7e3f72a25aa30cbb2eecb49036e0acf066 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Wed, 2 Aug 2023 12:55:46 +0200 Subject: x86/ibt: Suppress spurious ENDBR It was reported that under certain circumstances GCC emits ENDBR instructions for _THIS_IP_ usage. Specifically, when it appears at the start of a basic block -- but not elsewhere. Since _THIS_IP_ is never used for control flow, these ENDBR instructions are completely superfluous. Override the _THIS_IP_ definition for x86_64 to avoid this. Less ENDBR instructions is better. Fixes: 156ff4a544ae ("x86/ibt: Base IBT bits") Reported-by: David Kaplan Reviewed-by: Andrew Cooper Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://lore.kernel.org/r/20230802110323.016197440@infradead.org --- arch/x86/include/asm/linkage.h | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'arch') diff --git a/arch/x86/include/asm/linkage.h b/arch/x86/include/asm/linkage.h index 97a3de7892d3..5ff49fd67732 100644 --- a/arch/x86/include/asm/linkage.h +++ b/arch/x86/include/asm/linkage.h @@ -8,6 +8,14 @@ #undef notrace #define notrace __attribute__((no_instrument_function)) +#ifdef CONFIG_64BIT +/* + * The generic version tends to create spurious ENDBR instructions under + * certain conditions. + */ +#define _THIS_IP_ ({ unsigned long __here; asm ("lea 0(%%rip), %0" : "=r" (__here)); __here; }) +#endif + #ifdef CONFIG_X86_32 #define asmlinkage CPP_ASMLINKAGE __attribute__((regparm(0))) #endif /* CONFIG_X86_32 */ -- cgit v1.2.3 From 7575e5a35267983dcbeb1e0d3a49d21ae3cf0b82 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Wed, 2 Aug 2023 12:55:47 +0200 Subject: x86/ibt: Avoid duplicate ENDBR in __put_user_nocheck*() Commit cb855971d717 ("x86/putuser: Provide room for padding") changed __put_user_nocheck_*() into proper functions but failed to note that SYM_FUNC_START() already provides ENDBR, rendering the explicit ENDBR superfluous. Fixes: cb855971d717 ("x86/putuser: Provide room for padding") Reported-by: David Kaplan Reviewed-by: Andrew Cooper Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://lore.kernel.org/r/20230802110323.086971726@infradead.org --- arch/x86/lib/putuser.S | 4 ---- 1 file changed, 4 deletions(-) (limited to 'arch') diff --git a/arch/x86/lib/putuser.S b/arch/x86/lib/putuser.S index 1451e0c4ae22..235bbda6fc82 100644 --- a/arch/x86/lib/putuser.S +++ b/arch/x86/lib/putuser.S @@ -56,7 +56,6 @@ SYM_FUNC_END(__put_user_1) EXPORT_SYMBOL(__put_user_1) SYM_FUNC_START(__put_user_nocheck_1) - ENDBR ASM_STAC 2: movb %al,(%_ASM_CX) xor %ecx,%ecx @@ -76,7 +75,6 @@ SYM_FUNC_END(__put_user_2) EXPORT_SYMBOL(__put_user_2) SYM_FUNC_START(__put_user_nocheck_2) - ENDBR ASM_STAC 4: movw %ax,(%_ASM_CX) xor %ecx,%ecx @@ -96,7 +94,6 @@ SYM_FUNC_END(__put_user_4) EXPORT_SYMBOL(__put_user_4) SYM_FUNC_START(__put_user_nocheck_4) - ENDBR ASM_STAC 6: movl %eax,(%_ASM_CX) xor %ecx,%ecx @@ -119,7 +116,6 @@ SYM_FUNC_END(__put_user_8) EXPORT_SYMBOL(__put_user_8) SYM_FUNC_START(__put_user_nocheck_8) - ENDBR ASM_STAC 9: mov %_ASM_AX,(%_ASM_CX) #ifdef CONFIG_X86_32 -- cgit v1.2.3 From 1bfb2b618d52e59a4ef1896b46c4698ad2be66b7 Mon Sep 17 00:00:00 2001 From: Song Shuai Date: Wed, 6 Sep 2023 17:58:17 +0800 Subject: riscv: kexec: Align the kexeced kernel entry The current riscv boot protocol requires 2MB alignment for RV64 and 4MB alignment for RV32. In KEXEC_FILE path, the elf_find_pbase() function should align the kexeced kernel entry according to the requirement, otherwise the kexeced kernel would silently BUG at the setup_vm(). Fixes: 8acea455fafa ("RISC-V: Support for kexec_file on panic") Signed-off-by: Song Shuai Link: https://lore.kernel.org/r/20230906095817.364390-1-songshuaishuai@tinylab.org Signed-off-by: Palmer Dabbelt --- arch/riscv/kernel/elf_kexec.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) (limited to 'arch') diff --git a/arch/riscv/kernel/elf_kexec.c b/arch/riscv/kernel/elf_kexec.c index f4099059ed8f..e60fbd8660c4 100644 --- a/arch/riscv/kernel/elf_kexec.c +++ b/arch/riscv/kernel/elf_kexec.c @@ -98,7 +98,13 @@ static int elf_find_pbase(struct kimage *image, unsigned long kernel_len, kbuf.image = image; kbuf.buf_min = lowest_paddr; kbuf.buf_max = ULONG_MAX; - kbuf.buf_align = PAGE_SIZE; + + /* + * Current riscv boot protocol requires 2MB alignment for + * RV64 and 4MB alignment for RV32 + * + */ + kbuf.buf_align = PMD_SIZE; kbuf.mem = KEXEC_BUF_MEM_UNKNOWN; kbuf.memsz = ALIGN(kernel_len, PAGE_SIZE); kbuf.top_down = false; -- cgit v1.2.3 From 8eb8fe67e2c84324398f5983c41b4f831d0705b3 Mon Sep 17 00:00:00 2001 From: Icenowy Zheng Date: Tue, 12 Sep 2023 15:24:10 +0800 Subject: riscv: errata: fix T-Head dcache.cva encoding The dcache.cva encoding shown in the comments are wrong, it's for dcache.cval1 (which is restricted to L1) instead. Fix this in the comment and in the hardcoded instruction. Signed-off-by: Icenowy Zheng Tested-by: Sergey Matyukevich Reviewed-by: Heiko Stuebner Reviewed-by: Guo Ren Tested-by: Drew Fustini Link: https://lore.kernel.org/r/20230912072410.2481-1-jszhang@kernel.org Signed-off-by: Palmer Dabbelt --- arch/riscv/include/asm/errata_list.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch') diff --git a/arch/riscv/include/asm/errata_list.h b/arch/riscv/include/asm/errata_list.h index e2ecd01bfac7..b55b434f0059 100644 --- a/arch/riscv/include/asm/errata_list.h +++ b/arch/riscv/include/asm/errata_list.h @@ -105,7 +105,7 @@ asm volatile(ALTERNATIVE( \ * | 31 - 25 | 24 - 20 | 19 - 15 | 14 - 12 | 11 - 7 | 6 - 0 | * 0000001 01001 rs1 000 00000 0001011 * dcache.cva rs1 (clean, virtual address) - * 0000001 00100 rs1 000 00000 0001011 + * 0000001 00101 rs1 000 00000 0001011 * * dcache.cipa rs1 (clean then invalidate, physical address) * | 31 - 25 | 24 - 20 | 19 - 15 | 14 - 12 | 11 - 7 | 6 - 0 | @@ -118,7 +118,7 @@ asm volatile(ALTERNATIVE( \ * 0000000 11001 00000 000 00000 0001011 */ #define THEAD_inval_A0 ".long 0x0265000b" -#define THEAD_clean_A0 ".long 0x0245000b" +#define THEAD_clean_A0 ".long 0x0255000b" #define THEAD_flush_A0 ".long 0x0275000b" #define THEAD_SYNC_S ".long 0x0190000b" -- cgit v1.2.3 From 108af4b4bd3813610701379a58538e3339b162e4 Mon Sep 17 00:00:00 2001 From: Ricardo Neri Date: Mon, 14 Aug 2023 20:57:47 -0700 Subject: x86/sched: Restore the SD_ASYM_PACKING flag in the DIE domain Commit 8f2d6c41e5a6 ("x86/sched: Rewrite topology setup") dropped the SD_ASYM_PACKING flag in the DIE domain added in commit 044f0e27dec6 ("x86/sched: Add the SD_ASYM_PACKING flag to the die domain of hybrid processors"). Restore it on hybrid processors. The die-level domain does not depend on any build configuration and now x86_sched_itmt_flags() is always needed. Remove the build dependency on CONFIG_SCHED_[SMT|CLUSTER|MC]. Fixes: 8f2d6c41e5a6 ("x86/sched: Rewrite topology setup") Signed-off-by: Ricardo Neri Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Chen Yu Tested-by: Caleb Callaway Link: https://lkml.kernel.org/r/20230815035747.11529-1-ricardo.neri-calderon@linux.intel.com --- arch/x86/kernel/smpboot.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) (limited to 'arch') diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index d40ed3a7dc23..266d05e22ac3 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -579,7 +579,6 @@ static bool match_llc(struct cpuinfo_x86 *c, struct cpuinfo_x86 *o) } -#if defined(CONFIG_SCHED_SMT) || defined(CONFIG_SCHED_CLUSTER) || defined(CONFIG_SCHED_MC) static inline int x86_sched_itmt_flags(void) { return sysctl_sched_itmt_enabled ? SD_ASYM_PACKING : 0; @@ -603,7 +602,14 @@ static int x86_cluster_flags(void) return cpu_cluster_flags() | x86_sched_itmt_flags(); } #endif -#endif + +static int x86_die_flags(void) +{ + if (cpu_feature_enabled(X86_FEATURE_HYBRID_CPU)) + return x86_sched_itmt_flags(); + + return 0; +} /* * Set if a package/die has multiple NUMA nodes inside. @@ -640,7 +646,7 @@ static void __init build_sched_topology(void) */ if (!x86_has_numa_in_package) { x86_topology[i++] = (struct sched_domain_topology_level){ - cpu_cpu_mask, SD_INIT_NAME(DIE) + cpu_cpu_mask, x86_die_flags, SD_INIT_NAME(DIE) }; } -- cgit v1.2.3 From f530ee95b72e77b09c141c4b1a4b94d1199ffbd9 Mon Sep 17 00:00:00 2001 From: "Kirill A. Shutemov" Date: Fri, 15 Sep 2023 10:02:21 +0300 Subject: x86/boot/compressed: Reserve more memory for page tables The decompressor has a hard limit on the number of page tables it can allocate. This limit is defined at compile-time and will cause boot failure if it is reached. The kernel is very strict and calculates the limit precisely for the worst-case scenario based on the current configuration. However, it is easy to forget to adjust the limit when a new use-case arises. The worst-case scenario is rarely encountered during sanity checks. In the case of enabling 5-level paging, a use-case was overlooked. The limit needs to be increased by one to accommodate the additional level. This oversight went unnoticed until Aaron attempted to run the kernel via kexec with 5-level paging and unaccepted memory enabled. Update wost-case calculations to include 5-level paging. To address this issue, let's allocate some extra space for page tables. 128K should be sufficient for any use-case. The logic can be simplified by using a single value for all kernel configurations. [ Also add a warning, should this memory run low - by Dave Hansen. ] Fixes: 34bbb0009f3b ("x86/boot/compressed: Enable 5-level paging during decompression stage") Reported-by: Aaron Lu Signed-off-by: Kirill A. Shutemov Signed-off-by: Ingo Molnar Link: https://lore.kernel.org/r/20230915070221.10266-1-kirill.shutemov@linux.intel.com --- arch/x86/boot/compressed/ident_map_64.c | 8 ++++++ arch/x86/include/asm/boot.h | 45 +++++++++++++++++++++++---------- 2 files changed, 39 insertions(+), 14 deletions(-) (limited to 'arch') diff --git a/arch/x86/boot/compressed/ident_map_64.c b/arch/x86/boot/compressed/ident_map_64.c index bcc956c17872..08f93b0401bb 100644 --- a/arch/x86/boot/compressed/ident_map_64.c +++ b/arch/x86/boot/compressed/ident_map_64.c @@ -59,6 +59,14 @@ static void *alloc_pgt_page(void *context) return NULL; } + /* Consumed more tables than expected? */ + if (pages->pgt_buf_offset == BOOT_PGT_SIZE_WARN) { + debug_putstr("pgt_buf running low in " __FILE__ "\n"); + debug_putstr("Need to raise BOOT_PGT_SIZE?\n"); + debug_putaddr(pages->pgt_buf_offset); + debug_putaddr(pages->pgt_buf_size); + } + entry = pages->pgt_buf + pages->pgt_buf_offset; pages->pgt_buf_offset += PAGE_SIZE; diff --git a/arch/x86/include/asm/boot.h b/arch/x86/include/asm/boot.h index 4ae14339cb8c..b3a7cfb0d99e 100644 --- a/arch/x86/include/asm/boot.h +++ b/arch/x86/include/asm/boot.h @@ -40,23 +40,40 @@ #ifdef CONFIG_X86_64 # define BOOT_STACK_SIZE 0x4000 +/* + * Used by decompressor's startup_32() to allocate page tables for identity + * mapping of the 4G of RAM in 4-level paging mode: + * - 1 level4 table; + * - 1 level3 table; + * - 4 level2 table that maps everything with 2M pages; + * + * The additional level5 table needed for 5-level paging is allocated from + * trampoline_32bit memory. + */ # define BOOT_INIT_PGT_SIZE (6*4096) -# ifdef CONFIG_RANDOMIZE_BASE + /* - * Assuming all cross the 512GB boundary: - * 1 page for level4 - * (2+2)*4 pages for kernel, param, cmd_line, and randomized kernel - * 2 pages for first 2M (video RAM: CONFIG_X86_VERBOSE_BOOTUP). - * Total is 19 pages. + * Total number of page tables kernel_add_identity_map() can allocate, + * including page tables consumed by startup_32(). + * + * Worst-case scenario: + * - 5-level paging needs 1 level5 table; + * - KASLR needs to map kernel, boot_params, cmdline and randomized kernel, + * assuming all of them cross 256T boundary: + * + 4*2 level4 table; + * + 4*2 level3 table; + * + 4*2 level2 table; + * - X86_VERBOSE_BOOTUP needs to map the first 2M (video RAM): + * + 1 level4 table; + * + 1 level3 table; + * + 1 level2 table; + * Total: 28 tables + * + * Add 4 spare table in case decompressor touches anything beyond what is + * accounted above. Warn if it happens. */ -# ifdef CONFIG_X86_VERBOSE_BOOTUP -# define BOOT_PGT_SIZE (19*4096) -# else /* !CONFIG_X86_VERBOSE_BOOTUP */ -# define BOOT_PGT_SIZE (17*4096) -# endif -# else /* !CONFIG_RANDOMIZE_BASE */ -# define BOOT_PGT_SIZE BOOT_INIT_PGT_SIZE -# endif +# define BOOT_PGT_SIZE_WARN (28*4096) +# define BOOT_PGT_SIZE (32*4096) #else /* !CONFIG_X86_64 */ # define BOOT_STACK_SIZE 0x1000 -- cgit v1.2.3 From 75b2f7e4c9e0fd750a5a27ca9736d1daa7a3762a Mon Sep 17 00:00:00 2001 From: Song Liu Date: Thu, 14 Sep 2023 10:01:38 -0700 Subject: x86/purgatory: Remove LTO flags -flto* implies -ffunction-sections. With LTO enabled, ld.lld generates multiple .text sections for purgatory.ro: $ readelf -S purgatory.ro | grep " .text" [ 1] .text PROGBITS 0000000000000000 00000040 [ 7] .text.purgatory PROGBITS 0000000000000000 000020e0 [ 9] .text.warn PROGBITS 0000000000000000 000021c0 [13] .text.sha256_upda PROGBITS 0000000000000000 000022f0 [15] .text.sha224_upda PROGBITS 0000000000000000 00002be0 [17] .text.sha256_fina PROGBITS 0000000000000000 00002bf0 [19] .text.sha224_fina PROGBITS 0000000000000000 00002cc0 This causes WARNING from kexec_purgatory_setup_sechdrs(): WARNING: CPU: 26 PID: 110894 at kernel/kexec_file.c:919 kexec_load_purgatory+0x37f/0x390 Fix this by disabling LTO for purgatory. [ AFAICT, x86 is the only arch that supports LTO and purgatory. ] We could also fix this with an explicit linker script to rejoin .text.* sections back into .text. However, given the benefit of LTOing purgatory is small, simply disable the production of more .text.* sections for now. Fixes: b33fff07e3e3 ("x86, build: allow LTO to be selected") Signed-off-by: Song Liu Signed-off-by: Ingo Molnar Reviewed-by: Nick Desaulniers Reviewed-by: Sami Tolvanen Link: https://lore.kernel.org/r/20230914170138.995606-1-song@kernel.org --- arch/x86/purgatory/Makefile | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'arch') diff --git a/arch/x86/purgatory/Makefile b/arch/x86/purgatory/Makefile index c2a29be35c01..08aa0f25f12a 100644 --- a/arch/x86/purgatory/Makefile +++ b/arch/x86/purgatory/Makefile @@ -19,6 +19,10 @@ CFLAGS_sha256.o := -D__DISABLE_EXPORTS -D__NO_FORTIFY # optimization flags. KBUILD_CFLAGS := $(filter-out -fprofile-sample-use=% -fprofile-use=%,$(KBUILD_CFLAGS)) +# When LTO is enabled, llvm emits many text sections, which is not supported +# by kexec. Remove -flto=* flags. +KBUILD_CFLAGS := $(filter-out $(CC_FLAGS_LTO),$(KBUILD_CFLAGS)) + # When linking purgatory.ro with -r unresolved symbols are not checked, # also link a purgatory.chk binary without -r to check for unresolved symbols. PURGATORY_LDFLAGS := -e purgatory_start -z nodefaultlib -- cgit v1.2.3 From ef4d48368587f27cb1f690691518889d8fb3510b Mon Sep 17 00:00:00 2001 From: Anup Patel Date: Mon, 18 Sep 2023 11:48:29 +0530 Subject: RISC-V: KVM: Fix KVM_GET_REG_LIST API for ISA_EXT registers The ISA_EXT registers to enabled/disable ISA extensions for VCPU are always available when underlying host has the corresponding ISA extension. The copy_isa_ext_reg_indices() called by the KVM_GET_REG_LIST API does not align with this expectation so let's fix it. Fixes: 031f9efafc08 ("KVM: riscv: Add KVM_GET_REG_LIST API support") Signed-off-by: Anup Patel Reviewed-by: Atish Patra Reviewed-by: Andrew Jones Signed-off-by: Anup Patel --- arch/riscv/kvm/vcpu_onereg.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch') diff --git a/arch/riscv/kvm/vcpu_onereg.c b/arch/riscv/kvm/vcpu_onereg.c index 1b7e9fa265cb..e7e833ced91b 100644 --- a/arch/riscv/kvm/vcpu_onereg.c +++ b/arch/riscv/kvm/vcpu_onereg.c @@ -842,7 +842,7 @@ static int copy_isa_ext_reg_indices(const struct kvm_vcpu *vcpu, u64 reg = KVM_REG_RISCV | size | KVM_REG_RISCV_ISA_EXT | i; isa_ext = kvm_isa_ext_arr[i]; - if (!__riscv_isa_extension_available(vcpu->arch.isa, isa_ext)) + if (!__riscv_isa_extension_available(NULL, isa_ext)) continue; if (uindices) { -- cgit v1.2.3 From 17f71a2a340f1dcd397a66110005722177d5927c Mon Sep 17 00:00:00 2001 From: Anup Patel Date: Mon, 18 Sep 2023 11:58:29 +0530 Subject: RISC-V: KVM: Fix riscv_vcpu_get_isa_ext_single() for missing extensions The riscv_vcpu_get_isa_ext_single() should fail with -ENOENT error when corresponding ISA extension is not available on the host. Fixes: e98b1085be79 ("RISC-V: KVM: Factor-out ONE_REG related code to its own source file") Signed-off-by: Anup Patel Reviewed-by: Atish Patra Reviewed-by: Andrew Jones Signed-off-by: Anup Patel --- arch/riscv/kvm/vcpu_onereg.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'arch') diff --git a/arch/riscv/kvm/vcpu_onereg.c b/arch/riscv/kvm/vcpu_onereg.c index e7e833ced91b..b7e0e03c69b1 100644 --- a/arch/riscv/kvm/vcpu_onereg.c +++ b/arch/riscv/kvm/vcpu_onereg.c @@ -460,8 +460,11 @@ static int riscv_vcpu_get_isa_ext_single(struct kvm_vcpu *vcpu, reg_num >= ARRAY_SIZE(kvm_isa_ext_arr)) return -ENOENT; - *reg_val = 0; host_isa_ext = kvm_isa_ext_arr[reg_num]; + if (!__riscv_isa_extension_available(NULL, host_isa_ext)) + return -ENOENT; + + *reg_val = 0; if (__riscv_isa_extension_available(vcpu->arch.isa, host_isa_ext)) *reg_val = 1; /* Mark the given extension as available */ -- cgit v1.2.3