kernel.git/arch/riscv, branch linux-5.7.y

riscv: Fixup static_obj() fail

2020-08-26T09:42:12Z

[ Upstream commit 6184358da0004c8fd940afda6c0a0fa4027dc911 ] When enable LOCKDEP, static_obj() will cause error. Because some __initdata static variables is before _stext: static int static_obj(const void *obj) { unsigned long start = (unsigned long) &_stext, end = (unsigned long) &_end, addr = (unsigned long) obj; /* * static variable? */ if ((addr >= start) && (addr < end)) return 1; [ 0.067192] INFO: trying to register non-static key. [ 0.067325] the code is fine but needs lockdep annotation. [ 0.067449] turning off the locking correctness validator. [ 0.067718] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 5.7.0-rc7-dirty #44 [ 0.067945] Call Trace: [ 0.068369] [] walk_stackframe+0x0/0xa4 [ 0.068506] [] show_stack+0x2a/0x34 [ 0.068631] [] dump_stack+0x94/0xca [ 0.068757] [] register_lock_class+0x5b8/0x5bc [ 0.068969] [] __lock_acquire+0x6c/0x1d5c [ 0.069101] [] lock_acquire+0xae/0x312 [ 0.069228] [] _raw_spin_lock_irqsave+0x40/0x5a [ 0.069357] [] complete+0x1e/0x50 [ 0.069479] [] rest_init+0x1b0/0x28a [ 0.069660] [] 0xffffffe0000016a2 [ 0.069779] [] 0xffffffe000001b84 [ 0.069953] [] 0xffffffe000001092 static __initdata DECLARE_COMPLETION(kthreadd_done); noinline void __ref rest_init(void) { ... complete(&kthreadd_done); Signed-off-by: Guo Ren Signed-off-by: Palmer Dabbelt Signed-off-by: Sasha Levin

riscv: Parse all memory blocks to remove unusable memory

2020-08-05T07:58:51Z

[ Upstream commit fa5a198359053c8e21dcc2b39c0e13871059bc9f ] Currently, maximum physical memory allowed is equal to -PAGE_OFFSET. That's why we remove any memory blocks spanning beyond that size. However, it is done only for memblock containing linux kernel which will not work if there are multiple memblocks. Process all memory blocks to figure out how much memory needs to be removed and remove at the end instead of updating the memblock list in place. Signed-off-by: Atish Patra Signed-off-by: Palmer Dabbelt Signed-off-by: Sasha Levin

RISC-V: Set maximum number of mapped pages correctly

2020-08-05T07:58:50Z

[ Upstream commit d0d8aae64566b753c4330fbd5944b88af035f299 ] Currently, maximum number of mapper pages are set to the pfn calculated from the memblock size of the memblock containing kernel. This will work until that memblock spans the entire memory. However, it will be set to a wrong value if there are multiple memblocks defined in kernel (e.g. with efi runtime services). Set the the maximum value to the pfn calculated from dram size. Signed-off-by: Atish Patra Signed-off-by: Palmer Dabbelt Signed-off-by: Sasha Levin

riscv: kasan: use local_tlb_flush_all() to avoid uninitialized __sbi_rfence

2020-08-05T07:58:49Z

[ Upstream commit 4cb699d0447be8e0906539f93cbe41e19598ee5a ] It fails to boot the v5.8-rc4 kernel with CONFIG_KASAN because kasan_init and kasan_early_init use uninitialized __sbi_rfence as executing the tlb_flush_all(). Actually, at this moment, only the CPU which is responsible for the system initialization enables the MMU. Other CPUs are parking at the .Lsecondary_start. Hence the tlb_flush_all() is able to be replaced by local_tlb_flush_all() to avoid using uninitialized __sbi_rfence. Signed-off-by: Vincent Chen Signed-off-by: Palmer Dabbelt Signed-off-by: Sasha Levin

RISC-V: Upgrade smp_mb__after_spinlock() to iorw,iorw

2020-07-29T08:19:54Z

[ Upstream commit 38b7c2a3ffb1fce8358ddc6006cfe5c038ff9963 ] While digging through the recent mmiowb preemption issue it came up that we aren't actually preventing IO from crossing a scheduling boundary. While it's a bit ugly to overload smp_mb__after_spinlock() with this behavior, it's what PowerPC is doing so there's some precedent. Signed-off-by: Palmer Dabbelt Signed-off-by: Sasha Levin

RISC-V: Do not rely on initrd_start/end computed during early dt parsing

2020-07-29T08:19:49Z

[ Upstream commit 4400231c8acc7e513204c8470c6d796ba47dc169 ] Currently, initrd_start/end are computed during early_init_dt_scan but used during arch_setup. We will get the following panic if initrd is used and CONFIG_DEBUG_VIRTUAL is turned on. [ 0.000000] ------------[ cut here ]------------ [ 0.000000] kernel BUG at arch/riscv/mm/physaddr.c:33! [ 0.000000] Kernel BUG [#1] [ 0.000000] Modules linked in: [ 0.000000] CPU: 0 PID: 0 Comm: swapper Not tainted 5.8.0-rc4-00015-ged0b226fed02 #886 [ 0.000000] epc: ffffffe0002058d2 ra : ffffffe0000053f0 sp : ffffffe001001f40 [ 0.000000] gp : ffffffe00106e250 tp : ffffffe001009d40 t0 : ffffffe00107ee28 [ 0.000000] t1 : 0000000000000000 t2 : ffffffe000a2e880 s0 : ffffffe001001f50 [ 0.000000] s1 : ffffffe0001383e8 a0 : ffffffe00c087e00 a1 : 0000000080200000 [ 0.000000] a2 : 00000000010bf000 a3 : ffffffe00106f3c8 a4 : ffffffe0010bf000 [ 0.000000] a5 : ffffffe000000000 a6 : 0000000000000006 a7 : 0000000000000001 [ 0.000000] s2 : ffffffe00106f068 s3 : ffffffe00106f070 s4 : 0000000080200000 [ 0.000000] s5 : 0000000082200000 s6 : 0000000000000000 s7 : 0000000000000000 [ 0.000000] s8 : 0000000080011010 s9 : 0000000080012700 s10: 0000000000000000 [ 0.000000] s11: 0000000000000000 t3 : 000000000001fe30 t4 : 000000000001fe30 [ 0.000000] t5 : 0000000000000000 t6 : ffffffe00107c471 [ 0.000000] status: 0000000000000100 badaddr: 0000000000000000 cause: 0000000000000003 [ 0.000000] random: get_random_bytes called from print_oops_end_marker+0x22/0x46 with crng_init=0 To avoid the error, initrd_start/end can be computed from phys_initrd_start/size in setup itself. It also improves the initrd placement by aligning the start and size with the page size. Fixes: 76d2a0493a17 ("RISC-V: Init and Halt Code") Signed-off-by: Atish Patra Signed-off-by: Palmer Dabbelt Signed-off-by: Sasha Levin

riscv: use 16KB kernel stack on 64-bit

2020-07-22T07:34:23Z

commit 0cac21b02ba5f3095fd2dcc77c26a25a0b2432ed upstream. With the current 8KB stack size there are frequent overflows in a 64-bit configuration. We may split IRQ stacks off in the future, but this fixes a number of issues right now. Signed-off-by: Andreas Schwab Reviewed-by: Anup Patel [Palmer: mention irqstack in the commit text] Fixes: 7db91e57a0ac ("RISC-V: Task implementation") Cc: stable@vger.kernel.org Signed-off-by: Palmer Dabbelt Signed-off-by: Greg Kroah-Hartman

RISC-V: Don't allow write+exec only page mapping request in mmap

2020-06-30T19:36:19Z

[ Upstream commit e0d17c842c0f824fd4df9f4688709fc6907201e1 ] As per the table 4.4 of version "20190608-Priv-MSU-Ratified" of the RISC-V instruction set manual[0], the PTE permission bit combination of "write+exec only" is reserved for future use. Hence, don't allow such mapping request in mmap call. An issue is been reported by David Abdurachmanov, that while running stress-ng with "sysbadaddr" argument, RCU stalls are observed on RISC-V specific kernel. This issue arises when the stress-sysbadaddr request for pages with "write+exec only" permission bits and then passes the address obtain from this mmap call to various system call. For the riscv kernel, the mmap call should fail for this particular combination of permission bits since it's not valid. [0]: http://dabbelt.com/~palmer/keep/riscv-isa-manual/riscv-privileged-20190608-1.pdf Signed-off-by: Yash Shah Reported-by: David Abdurachmanov [Palmer: Refer to the latest ISA specification at the only link I could find, and update the terminology.] Signed-off-by: Palmer Dabbelt Signed-off-by: Sasha Levin

riscv/atomic: Fix sign extension for RV64I

2020-06-30T19:36:14Z

[ Upstream commit 6c58f25e6938c073198af8b1e1832f83f8f0df33 ] The argument passed to cmpxchg is not guaranteed to be sign extended, but lr.w sign extends on RV64I. This makes cmpxchg fail on clang built kernels when __old is negative. To fix this, we just cast __old to long which sign extends on RV64I. With this fix, clang built RISC-V kernels now boot. Link: https://github.com/ClangBuiltLinux/linux/issues/867 Signed-off-by: Nathan Huckleberry Signed-off-by: Palmer Dabbelt Signed-off-by: Sasha Levin

RISC-V: Don't mark init section as non-executable

2020-06-22T07:33:08Z

commit 4e0f9e3a6104261f25b16fcab02fc96f5666ba11 upstream. The head text section (i.e. _start, secondary_start_sbi, etc) and the init section fall under same page table level-1 mapping. Currently, the runtime CPU hotplug is broken because we are marking init section as non-executable which in-turn marks head text section as non-executable. Further investigating other architectures, it seems marking the init section as non-executable is redundant because the init section pages are anyway poisoned and freed. To fix broken runtime CPU hotplug, we simply remove the code marking the init section as non-executable. Fixes: d27c3c90817e ("riscv: add STRICT_KERNEL_RWX support") Cc: stable@vger.kernel.org Signed-off-by: Anup Patel Reviewed-by: Zong Li Reviewed-by: Atish Patra Signed-off-by: Palmer Dabbelt Signed-off-by: Greg Kroah-Hartman