Messages in this thread | | | Date | Thu, 11 Aug 2022 10:09:42 -0700 (PDT) | Subject | Re: [PATCH] riscv/kprobes: allocate detour buffer from module area | From | Palmer Dabbelt <> |
| |
On Tue, 28 Jun 2022 18:13:17 PDT (-0700), liaochang1@huawei.com wrote: > To address the limitation of PC-relative branch instruction > on riscv architecture, detour buffer slot is allocated from > a area, the distance of which from kernel should be less than 4GB. > > For the time being, Modules region always live before the kernel. > But Vmalloc region reside far away from kernel, the distance is > half of the kernel address space. > > Signed-off-by: Liao Chang <liaochang1@huawei.com> > --- > arch/riscv/kernel/probes/kprobes.c | 20 ++++++++++++++++++++ > 1 file changed, 20 insertions(+) > > diff --git a/arch/riscv/kernel/probes/kprobes.c b/arch/riscv/kernel/probes/kprobes.c > index e6e950b7cf32..bc027a663b17 100644 > --- a/arch/riscv/kernel/probes/kprobes.c > +++ b/arch/riscv/kernel/probes/kprobes.c > @@ -6,12 +6,14 @@ > #include <linux/extable.h> > #include <linux/slab.h> > #include <linux/stop_machine.h> > +#include <linux/moduleloader.h> > #include <asm/ptrace.h> > #include <linux/uaccess.h> > #include <asm/sections.h> > #include <asm/cacheflush.h> > #include <asm/bug.h> > #include <asm/patch.h> > +#include <asm/set_memory.h> > > #include "decode-insn.h" > > @@ -86,10 +88,28 @@ int __kprobes arch_prepare_kprobe(struct kprobe *p) > #ifdef CONFIG_MMU > void *alloc_insn_page(void) > { > +#if defined(CONFIG_MODULES) && defined(CONFIG_64BIT) > + void *page; > + > + page = module_alloc(PAGE_SIZE); > + if (!page) > + return NULL; > + > + set_vm_flush_reset_perms(page); > + /* > + * First make the page read-only, and only then make it executable to > + * prevent it from being W+X in between. > + */ > + set_memory_ro((unsigned long)page, 1); > + set_memory_x((unsigned long)page, 1); > + > + return page; > +#else > return __vmalloc_node_range(PAGE_SIZE, 1, VMALLOC_START, VMALLOC_END, > GFP_KERNEL, PAGE_KERNEL_READ_EXEC, > VM_FLUSH_RESET_PERMS, NUMA_NO_NODE, > __builtin_return_address(0)); > +#endif > } > #endif
This probably shouldn't depend on CONFIG_MODULES, it's just as broken to assume we can jump to the vmalloc region regardless of whether modules are enabled or disabled. We should be able to just allocate from the jump-safe region either way.
| |