From d0e4dae74470fb709fc0ab61862c317938f4cc4d Mon Sep 17 00:00:00 2001 From: Bin Meng Date: Sun, 27 Jun 2021 21:51:17 +0800 Subject: riscv: Fix 32-bit RISC-V boot failure Commit dd2d082b5760 ("riscv: Cleanup setup_bootmem()") adjusted the calling sequence in setup_bootmem(), which invalidates the fix commit de043da0b9e7 ("RISC-V: Fix usage of memblock_enforce_memory_limit") did for 32-bit RISC-V unfortunately. So now 32-bit RISC-V does not boot again when testing booting kernel on QEMU 'virt' with '-m 2G', which was exactly what the original commit de043da0b9e7 ("RISC-V: Fix usage of memblock_enforce_memory_limit") tried to fix. Fixes: dd2d082b5760 ("riscv: Cleanup setup_bootmem()") Signed-off-by: Bin Meng Signed-off-by: Palmer Dabbelt --- arch/riscv/mm/init.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'arch') diff --git a/arch/riscv/mm/init.c b/arch/riscv/mm/init.c index 4faf8bd157ea..0fc72603c699 100644 --- a/arch/riscv/mm/init.c +++ b/arch/riscv/mm/init.c @@ -123,7 +123,7 @@ void __init setup_bootmem(void) { phys_addr_t vmlinux_end = __pa_symbol(&_end); phys_addr_t vmlinux_start = __pa_symbol(&_start); - phys_addr_t dram_end = memblock_end_of_DRAM(); + phys_addr_t dram_end; phys_addr_t max_mapped_addr = __pa(~(ulong)0); #ifdef CONFIG_XIP_KERNEL @@ -146,6 +146,8 @@ void __init setup_bootmem(void) #endif memblock_reserve(vmlinux_start, vmlinux_end - vmlinux_start); + dram_end = memblock_end_of_DRAM(); + /* * memblock allocator is not aware of the fact that last 4K bytes of * the addressable memory can not be mapped because of IS_ERR_VALUE -- cgit v1.2.3 From c79e89ecaa246c880292ba68cbe08c9c30db77e3 Mon Sep 17 00:00:00 2001 From: Heinrich Schuchardt Date: Tue, 29 Jun 2021 15:40:18 +0200 Subject: RISC-V: load initrd wherever it fits into memory Requiring that initrd is loaded below RAM start + 256 MiB led to failure to boot SUSE Linux with GRUB on QEMU, cf. https://lists.gnu.org/archive/html/grub-devel/2021-06/msg00037.html Remove the constraint. Reported-by: Andreas Schwab Signed-off-by: Heinrich Schuchardt Reviewed-by: Atish Patra Acked-by: Ard Biesheuvel Fixes: d7071743db31 ("RISC-V: Add EFI stub support.") Signed-off-by: Palmer Dabbelt --- arch/riscv/include/asm/efi.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch') diff --git a/arch/riscv/include/asm/efi.h b/arch/riscv/include/asm/efi.h index 6d98cd999680..7b3483ba2e84 100644 --- a/arch/riscv/include/asm/efi.h +++ b/arch/riscv/include/asm/efi.h @@ -27,10 +27,10 @@ int efi_set_mapping_permissions(struct mm_struct *mm, efi_memory_desc_t *md); #define ARCH_EFI_IRQ_FLAGS_MASK (SR_IE | SR_SPIE) -/* Load initrd at enough distance from DRAM start */ +/* Load initrd anywhere in system RAM */ static inline unsigned long efi_get_max_initrd_addr(unsigned long image_addr) { - return image_addr + SZ_256M; + return ULONG_MAX; } #define alloc_screen_info(x...) (&screen_info) -- cgit v1.2.3 From c09dc9e1cd3c205f66b2505d742e819735c0eb6f Mon Sep 17 00:00:00 2001 From: Alexandre Ghiti Date: Tue, 29 Jun 2021 11:13:46 +0200 Subject: riscv: Fix memory_limit for 64-bit kernel As described in Documentation/riscv/vm-layout.rst, the end of the virtual address space for 64-bit kernel is occupied by the modules/BPF/ kernel mappings so this actually reduces the amount of memory we are able to map and then use in the linear mapping. So make sure this limit is correctly set. Signed-off-by: Alexandre Ghiti Fixes: 2bfc6cd81bd1 ("riscv: Move kernel mapping outside of linear mapping") Signed-off-by: Palmer Dabbelt --- arch/riscv/mm/init.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) (limited to 'arch') diff --git a/arch/riscv/mm/init.c b/arch/riscv/mm/init.c index ac48742fa6fc..e4356d65fdce 100644 --- a/arch/riscv/mm/init.c +++ b/arch/riscv/mm/init.c @@ -127,10 +127,17 @@ void __init mem_init(void) } /* - * The default maximal physical memory size is -PAGE_OFFSET, - * limit the memory size via mem. + * The default maximal physical memory size is -PAGE_OFFSET for 32-bit kernel, + * whereas for 64-bit kernel, the end of the virtual address space is occupied + * by the modules/BPF/kernel mappings which reduces the available size of the + * linear mapping. + * Limit the memory size via mem. */ +#ifdef CONFIG_64BIT +static phys_addr_t memory_limit = -PAGE_OFFSET - SZ_4G; +#else static phys_addr_t memory_limit = -PAGE_OFFSET; +#endif static int __init early_mem(char *p) { -- cgit v1.2.3 From c99127c452484ac89c75af39c3c865174ce2de99 Mon Sep 17 00:00:00 2001 From: Alexandre Ghiti Date: Tue, 29 Jun 2021 11:13:47 +0200 Subject: riscv: Make sure the linear mapping does not use the kernel mapping For 64-bit kernel, the end of the address space is occupied by the kernel mapping and currently, the functions to populate the kernel page tables (i.e. create_p*d_mapping) do not override existing mapping so we must make sure the linear mapping does not map memory in the kernel mapping by clipping the memory above the memory limit. Signed-off-by: Alexandre Ghiti Fixes: c9811e379b21 ("riscv: Add mem kernel parameter support") Signed-off-by: Palmer Dabbelt --- arch/riscv/mm/init.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch') diff --git a/arch/riscv/mm/init.c b/arch/riscv/mm/init.c index e4356d65fdce..644a34b0d77d 100644 --- a/arch/riscv/mm/init.c +++ b/arch/riscv/mm/init.c @@ -717,6 +717,8 @@ static void __init setup_vm_final(void) if (start <= __pa(PAGE_OFFSET) && __pa(PAGE_OFFSET) < end) start = __pa(PAGE_OFFSET); + if (end >= __pa(PAGE_OFFSET) + memory_limit) + end = __pa(PAGE_OFFSET) + memory_limit; map_size = best_map_size(start, end - start); for (pa = start; pa < end; pa += map_size) { -- cgit v1.2.3 From db6b84a368b495cb7e41be9cb9e73d4d0537d027 Mon Sep 17 00:00:00 2001 From: Alexandre Ghiti Date: Tue, 29 Jun 2021 11:13:48 +0200 Subject: riscv: Make sure the kernel mapping does not overlap with IS_ERR_VALUE The check that is done in setup_bootmem currently only works for 32-bit kernel since the kernel mapping has been moved outside of the linear mapping for 64-bit kernel. So make sure that for 64-bit kernel, the kernel mapping does not overlap with the last 4K of the addressable memory. Signed-off-by: Alexandre Ghiti Fixes: 2bfc6cd81bd1 ("riscv: Move kernel mapping outside of linear mapping") Signed-off-by: Palmer Dabbelt --- arch/riscv/mm/init.c | 18 ++++++++++++++++-- 1 file changed, 16 insertions(+), 2 deletions(-) (limited to 'arch') diff --git a/arch/riscv/mm/init.c b/arch/riscv/mm/init.c index 644a34b0d77d..a14bf3910eec 100644 --- a/arch/riscv/mm/init.c +++ b/arch/riscv/mm/init.c @@ -159,7 +159,7 @@ static void __init setup_bootmem(void) { phys_addr_t vmlinux_end = __pa_symbol(&_end); phys_addr_t vmlinux_start = __pa_symbol(&_start); - phys_addr_t max_mapped_addr = __pa(~(ulong)0); + phys_addr_t __maybe_unused max_mapped_addr; phys_addr_t dram_end; #ifdef CONFIG_XIP_KERNEL @@ -183,14 +183,20 @@ static void __init setup_bootmem(void) dram_end = memblock_end_of_DRAM(); +#ifndef CONFIG_64BIT /* * memblock allocator is not aware of the fact that last 4K bytes of * the addressable memory can not be mapped because of IS_ERR_VALUE * macro. Make sure that last 4k bytes are not usable by memblock - * if end of dram is equal to maximum addressable memory. + * if end of dram is equal to maximum addressable memory. For 64-bit + * kernel, this problem can't happen here as the end of the virtual + * address space is occupied by the kernel mapping then this check must + * be done in create_kernel_page_table. */ + max_mapped_addr = __pa(~(ulong)0); if (max_mapped_addr == (dram_end - 1)) memblock_set_current_limit(max_mapped_addr - 4096); +#endif min_low_pfn = PFN_UP(memblock_start_of_DRAM()); max_low_pfn = max_pfn = PFN_DOWN(dram_end); @@ -578,6 +584,14 @@ asmlinkage void __init setup_vm(uintptr_t dtb_pa) BUG_ON((PAGE_OFFSET % PGDIR_SIZE) != 0); BUG_ON((kernel_map.phys_addr % map_size) != 0); +#ifdef CONFIG_64BIT + /* + * The last 4K bytes of the addressable memory can not be mapped because + * of IS_ERR_VALUE macro. + */ + BUG_ON((kernel_map.virt_addr + kernel_map.size) > ADDRESS_SPACE_END - SZ_4K); +#endif + pt_ops.alloc_pte = alloc_pte_early; pt_ops.get_pte_virt = get_pte_virt_early; #ifndef __PAGETABLE_PMD_FOLDED -- cgit v1.2.3 From 76f5dfacfb42b75e5782c017827877cfcee20474 Mon Sep 17 00:00:00 2001 From: Jisheng Zhang Date: Fri, 23 Jul 2021 08:22:26 +0800 Subject: riscv: stacktrace: pin the task's stack in get_wchan Pin the task's stack before calling walk_stackframe() in get_wchan(). This can fix the panic as reported by Andreas when CONFIG_VMAP_STACK=y: [ 65.609696] Unable to handle kernel paging request at virtual address ffffffd0003bbde8 [ 65.610460] Oops [#1] [ 65.610626] Modules linked in: virtio_blk virtio_mmio rtc_goldfish btrfs blake2b_generic libcrc32c xor raid6_pq sg dm_multipath dm_mod scsi_dh_rdac scsi_dh_emc scsi_dh_alua efivarfs [ 65.611670] CPU: 2 PID: 1 Comm: systemd Not tainted 5.14.0-rc1-1.g34fe32a-default #1 openSUSE Tumbleweed (unreleased) c62f7109153e5a0897ee58ba52393ad99b070fd2 [ 65.612334] Hardware name: riscv-virtio,qemu (DT) [ 65.613008] epc : get_wchan+0x5c/0x88 [ 65.613334] ra : get_wchan+0x42/0x88 [ 65.613625] epc : ffffffff800048a4 ra : ffffffff8000488a sp : ffffffd00021bb90 [ 65.614008] gp : ffffffff817709f8 tp : ffffffe07fe91b80 t0 : 00000000000001f8 [ 65.614411] t1 : 0000000000020000 t2 : 0000000000000000 s0 : ffffffd00021bbd0 [ 65.614818] s1 : ffffffd0003bbdf0 a0 : 0000000000000001 a1 : 0000000000000002 [ 65.615237] a2 : ffffffff81618008 a3 : 0000000000000000 a4 : 0000000000000000 [ 65.615637] a5 : ffffffd0003bc000 a6 : 0000000000000002 a7 : ffffffe27d370000 [ 65.616022] s2 : ffffffd0003bbd90 s3 : ffffffff8071a81e s4 : 0000000000003fff [ 65.616407] s5 : ffffffffffffc000 s6 : 0000000000000000 s7 : ffffffff81618008 [ 65.616845] s8 : 0000000000000001 s9 : 0000000180000040 s10: 0000000000000000 [ 65.617248] s11: 000000000000016b t3 : 000000ff00000000 t4 : 0c6aec92de5e3fd7 [ 65.617672] t5 : fff78f60608fcfff t6 : 0000000000000078 [ 65.618088] status: 0000000000000120 badaddr: ffffffd0003bbde8 cause: 000000000000000d [ 65.618621] [] get_wchan+0x5c/0x88 [ 65.619008] [] do_task_stat+0x7a2/0xa46 [ 65.619325] [] proc_tgid_stat+0xe/0x16 [ 65.619637] [] proc_single_show+0x46/0x96 [ 65.619979] [] seq_read_iter+0x190/0x31e [ 65.620341] [] seq_read+0xc4/0x104 [ 65.620633] [] vfs_read+0x6a/0x112 [ 65.620922] [] ksys_read+0x54/0xbe [ 65.621206] [] sys_read+0xe/0x16 [ 65.621474] [] ret_from_syscall+0x0/0x2 [ 65.622169] ---[ end trace f24856ed2b8789c5 ]--- [ 65.622832] Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b Signed-off-by: Jisheng Zhang Signed-off-by: Palmer Dabbelt --- arch/riscv/kernel/stacktrace.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) (limited to 'arch') diff --git a/arch/riscv/kernel/stacktrace.c b/arch/riscv/kernel/stacktrace.c index ff467b98c3e3..ac7593607fa6 100644 --- a/arch/riscv/kernel/stacktrace.c +++ b/arch/riscv/kernel/stacktrace.c @@ -132,8 +132,12 @@ unsigned long get_wchan(struct task_struct *task) { unsigned long pc = 0; - if (likely(task && task != current && !task_is_running(task))) + if (likely(task && task != current && !task_is_running(task))) { + if (!try_get_task_stack(task)) + return 0; walk_stackframe(task, NULL, save_wchan, &pc); + put_task_stack(task); + } return pc; } -- cgit v1.2.3 From 6010d300f9f7e16d1bf327b4730bcd0c0886d9e6 Mon Sep 17 00:00:00 2001 From: Akira Tsukamoto Date: Tue, 20 Jul 2021 17:50:52 +0900 Subject: riscv: __asm_copy_to-from_user: Fix: overrun copy There were two causes for the overrun memory access. The threshold size was too small. The aligning dst require one SZREG and unrolling word copy requires 8*SZREG, total have to be at least 9*SZREG. Inside the unrolling copy, the subtracting -(8*SZREG-1) would make iteration happening one extra loop. Proper value is -(8*SZREG). Signed-off-by: Akira Tsukamoto Fixes: ca6eaaa210de ("riscv: __asm_copy_to-from_user: Optimize unaligned memory access and pipeline stall") Signed-off-by: Palmer Dabbelt --- arch/riscv/lib/uaccess.S | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'arch') diff --git a/arch/riscv/lib/uaccess.S b/arch/riscv/lib/uaccess.S index bceb0629e440..8bbeca89a93f 100644 --- a/arch/riscv/lib/uaccess.S +++ b/arch/riscv/lib/uaccess.S @@ -35,7 +35,7 @@ ENTRY(__asm_copy_from_user) /* * Use byte copy only if too small. */ - li a3, 8*SZREG /* size must be larger than size in word_copy */ + li a3, 9*SZREG /* size must be larger than size in word_copy */ bltu a2, a3, .Lbyte_copy_tail /* @@ -75,7 +75,7 @@ ENTRY(__asm_copy_from_user) * a3 - a1 & mask:(SZREG-1) * t0 - end of aligned dst */ - addi t0, t0, -(8*SZREG-1) /* not to over run */ + addi t0, t0, -(8*SZREG) /* not to over run */ 2: fixup REG_L a4, 0(a1), 10f fixup REG_L a5, SZREG(a1), 10f @@ -97,7 +97,7 @@ ENTRY(__asm_copy_from_user) addi a1, a1, 8*SZREG bltu a0, t0, 2b - addi t0, t0, 8*SZREG-1 /* revert to original value */ + addi t0, t0, 8*SZREG /* revert to original value */ j .Lbyte_copy_tail .Lshift_copy: -- cgit v1.2.3 From 22b5f16ffeff38938ad7420a2bfa3c281c36fd17 Mon Sep 17 00:00:00 2001 From: Akira Tsukamoto Date: Tue, 20 Jul 2021 17:51:45 +0900 Subject: riscv: __asm_copy_to-from_user: Fix: fail on RV32 Had a bug when converting bytes to bits when the cpu was rv32. The a3 contains the number of bytes and multiple of 8 would be the bits. The LGREG is holding 2 for RV32 and 3 for RV32, so to achieve multiple of 8 it must always be constant 3. The 2 was mistakenly used for rv32. Signed-off-by: Akira Tsukamoto Fixes: ca6eaaa210de ("riscv: __asm_copy_to-from_user: Optimize unaligned memory access and pipeline stall") Signed-off-by: Palmer Dabbelt --- arch/riscv/lib/uaccess.S | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch') diff --git a/arch/riscv/lib/uaccess.S b/arch/riscv/lib/uaccess.S index 8bbeca89a93f..279876821969 100644 --- a/arch/riscv/lib/uaccess.S +++ b/arch/riscv/lib/uaccess.S @@ -125,7 +125,7 @@ ENTRY(__asm_copy_from_user) * t3 - prev shift * t4 - current shift */ - slli t3, a3, LGREG + slli t3, a3, 3 /* converting bytes in a3 to bits */ li a5, SZREG*8 sub t4, a5, t3 -- cgit v1.2.3 From d4b3e0105e3c2411af666a50b1bf2d25656a5e83 Mon Sep 17 00:00:00 2001 From: Akira Tsukamoto Date: Tue, 20 Jul 2021 17:52:36 +0900 Subject: riscv: __asm_copy_to-from_user: Remove unnecessary size check Clean up: The size of 0 will be evaluated in the next step. Not required here. Signed-off-by: Akira Tsukamoto Fixes: ca6eaaa210de ("riscv: __asm_copy_to-from_user: Optimize unaligned memory access and pipeline stall") Signed-off-by: Palmer Dabbelt --- arch/riscv/lib/uaccess.S | 1 - 1 file changed, 1 deletion(-) (limited to 'arch') diff --git a/arch/riscv/lib/uaccess.S b/arch/riscv/lib/uaccess.S index 279876821969..54d497a03164 100644 --- a/arch/riscv/lib/uaccess.S +++ b/arch/riscv/lib/uaccess.S @@ -30,7 +30,6 @@ ENTRY(__asm_copy_from_user) * t0 - end of uncopied dst */ add t0, a0, a2 - bgtu a0, t0, 5f /* * Use byte copy only if too small. -- cgit v1.2.3 From ea196c548c0ac407afd31d142712b6da8bd00244 Mon Sep 17 00:00:00 2001 From: Akira Tsukamoto Date: Tue, 20 Jul 2021 17:53:23 +0900 Subject: riscv: __asm_copy_to-from_user: Fix: Typos in comments Fixing typos and grammar mistakes and using more intuitive label name. Signed-off-by: Akira Tsukamoto Fixes: ca6eaaa210de ("riscv: __asm_copy_to-from_user: Optimize unaligned memory access and pipeline stall") Signed-off-by: Palmer Dabbelt --- arch/riscv/lib/uaccess.S | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-) (limited to 'arch') diff --git a/arch/riscv/lib/uaccess.S b/arch/riscv/lib/uaccess.S index 54d497a03164..63bc691cff91 100644 --- a/arch/riscv/lib/uaccess.S +++ b/arch/riscv/lib/uaccess.S @@ -33,19 +33,20 @@ ENTRY(__asm_copy_from_user) /* * Use byte copy only if too small. + * SZREG holds 4 for RV32 and 8 for RV64 */ li a3, 9*SZREG /* size must be larger than size in word_copy */ bltu a2, a3, .Lbyte_copy_tail /* - * Copy first bytes until dst is align to word boundary. + * Copy first bytes until dst is aligned to word boundary. * a0 - start of dst * t1 - start of aligned dst */ addi t1, a0, SZREG-1 andi t1, t1, ~(SZREG-1) /* dst is already aligned, skip */ - beq a0, t1, .Lskip_first_bytes + beq a0, t1, .Lskip_align_dst 1: /* a5 - one byte for copying data */ fixup lb a5, 0(a1), 10f @@ -54,7 +55,7 @@ ENTRY(__asm_copy_from_user) addi a0, a0, 1 /* dst */ bltu a0, t1, 1b /* t1 - start of aligned dst */ -.Lskip_first_bytes: +.Lskip_align_dst: /* * Now dst is aligned. * Use shift-copy if src is misaligned. @@ -71,7 +72,6 @@ ENTRY(__asm_copy_from_user) * * a0 - start of aligned dst * a1 - start of aligned src - * a3 - a1 & mask:(SZREG-1) * t0 - end of aligned dst */ addi t0, t0, -(8*SZREG) /* not to over run */ @@ -106,7 +106,7 @@ ENTRY(__asm_copy_from_user) * For misaligned copy we still perform aligned word copy, but * we need to use the value fetched from the previous iteration and * do some shifts. - * This is safe because reading less than a word size. + * This is safe because reading is less than a word size. * * a0 - start of aligned dst * a1 - start of src @@ -116,7 +116,7 @@ ENTRY(__asm_copy_from_user) */ /* calculating aligned word boundary for dst */ andi t1, t0, ~(SZREG-1) - /* Converting unaligned src to aligned arc */ + /* Converting unaligned src to aligned src */ andi a1, a1, ~(SZREG-1) /* @@ -128,7 +128,7 @@ ENTRY(__asm_copy_from_user) li a5, SZREG*8 sub t4, a5, t3 - /* Load the first word to combine with seceond word */ + /* Load the first word to combine with second word */ fixup REG_L a5, 0(a1), 10f 3: @@ -160,7 +160,7 @@ ENTRY(__asm_copy_from_user) * a1 - start of remaining src * t0 - end of remaining dst */ - bgeu a0, t0, 5f + bgeu a0, t0, .Lout_copy_user /* check if end of copy */ 4: fixup lb a5, 0(a1), 10f addi a1, a1, 1 /* src */ @@ -168,7 +168,7 @@ ENTRY(__asm_copy_from_user) addi a0, a0, 1 /* dst */ bltu a0, t0, 4b /* t0 - end of dst */ -5: +.Lout_copy_user: /* Disable access to user memory */ csrc CSR_STATUS, t6 li a0, 0 -- cgit v1.2.3