Message ID | 20241211133403.208920-7-jolsa@kernel.org (mailing list archive) |
---|---|
State | Changes Requested |
Delegated to: | BPF |
Headers | show |
Series | uprobes: Add support to optimize usdt probes on x86_64 | expand |
On 2024-12-11 14:33:55+0100, Jiri Olsa wrote: > Adding new uprobe syscall that calls uprobe handlers for given > 'breakpoint' address. > > The idea is that the 'breakpoint' address calls the user space > trampoline which executes the uprobe syscall. > > The syscall handler reads the return address of the initial call > to retrieve the original 'breakpoint' address. With this address > we find the related uprobe object and call its consumers. > > Adding the arch_uprobe_trampoline_mapping function that provides > uprobe trampoline mapping. This mapping is backed with one global > page initialized at __init time and shared by the all the mapping > instances. > > We do not allow to execute uprobe syscall if the caller is not > from uprobe trampoline mapping. > > Signed-off-by: Jiri Olsa <jolsa@kernel.org> > --- > arch/x86/entry/syscalls/syscall_64.tbl | 1 + > arch/x86/kernel/uprobes.c | 80 ++++++++++++++++++++++++++ > include/linux/syscalls.h | 2 + > include/linux/uprobes.h | 1 + > kernel/events/uprobes.c | 22 +++++++ > kernel/sys_ni.c | 1 + > 6 files changed, 107 insertions(+) > > diff --git a/arch/x86/entry/syscalls/syscall_64.tbl b/arch/x86/entry/syscalls/syscall_64.tbl > index 5eb708bff1c7..88e388c7675b 100644 > --- a/arch/x86/entry/syscalls/syscall_64.tbl > +++ b/arch/x86/entry/syscalls/syscall_64.tbl > @@ -345,6 +345,7 @@ > 333 common io_pgetevents sys_io_pgetevents > 334 common rseq sys_rseq > 335 common uretprobe sys_uretprobe > +336 common uprobe sys_uprobe > # don't use numbers 387 through 423, add new calls after the last > # 'common' entry > 424 common pidfd_send_signal sys_pidfd_send_signal > diff --git a/arch/x86/kernel/uprobes.c b/arch/x86/kernel/uprobes.c > index 22a17c149a55..23e4f2821cff 100644 > --- a/arch/x86/kernel/uprobes.c > +++ b/arch/x86/kernel/uprobes.c > @@ -425,6 +425,86 @@ SYSCALL_DEFINE0(uretprobe) > return -1; > } > > +static int tramp_mremap(const struct vm_special_mapping *sm, struct vm_area_struct *new_vma) > +{ > + return -EPERM; > +} > + > +static struct vm_special_mapping tramp_mapping = { > + .name = "[uprobes-trampoline]", > + .mremap = tramp_mremap, > +}; > + > +SYSCALL_DEFINE0(uprobe) > +{ > + struct pt_regs *regs = task_pt_regs(current); > + struct vm_area_struct *vma; > + unsigned long bp_vaddr; > + int err; > + > + err = copy_from_user(&bp_vaddr, (void __user *)regs->sp + 3*8, sizeof(bp_vaddr)); A #define for the magic values would be nice. > + if (err) { > + force_sig(SIGILL); > + return -1; > + } > + > + /* Allow execution only from uprobe trampolines. */ > + vma = vma_lookup(current->mm, regs->ip); > + if (!vma || vma->vm_private_data != (void *) &tramp_mapping) { vma_is_special_mapping() > + force_sig(SIGILL); > + return -1; > + } > + > + handle_syscall_uprobe(regs, bp_vaddr - 5); > + return 0; > +} > + > +asm ( > + ".pushsection .rodata\n" > + ".global uprobe_trampoline_entry\n" > + "uprobe_trampoline_entry:\n" > + "endbr64\n" > + "push %rcx\n" > + "push %r11\n" > + "push %rax\n" > + "movq $" __stringify(__NR_uprobe) ", %rax\n" > + "syscall\n" > + "pop %rax\n" > + "pop %r11\n" > + "pop %rcx\n" > + "ret\n" > + ".global uprobe_trampoline_end\n" > + "uprobe_trampoline_end:\n" > + ".popsection\n" > +); > + > +extern __visible u8 uprobe_trampoline_entry[]; > +extern __visible u8 uprobe_trampoline_end[]; > + > +const struct vm_special_mapping *arch_uprobe_trampoline_mapping(void) > +{ > + struct pt_regs *regs = task_pt_regs(current); > + > + return user_64bit_mode(regs) ? &tramp_mapping : NULL; > +} > + > +static int __init arch_uprobes_init(void) > +{ > + unsigned long size = uprobe_trampoline_end - uprobe_trampoline_entry; > + static struct page *pages[2]; > + struct page *page; > + > + page = alloc_page(GFP_HIGHUSER); That page could be in static memory, removing the need for the explicit allocation. It could also be __ro_after_init. Then tramp_mapping itself can be const. Also this seems to waste the page on 32bit kernels. > + if (!page) > + return -ENOMEM; > + pages[0] = page; > + tramp_mapping.pages = (struct page **) &pages; tramp_mapping.pages = pages; ? > + arch_uprobe_copy_ixol(page, 0, uprobe_trampoline_entry, size); > + return 0; > +} > + > +late_initcall(arch_uprobes_init); > + > /* > * If arch_uprobe->insn doesn't use rip-relative addressing, return > * immediately. Otherwise, rewrite the instruction so that it accesses [..]
On Fri, Dec 13, 2024 at 02:48:00PM +0100, Thomas Weißschuh wrote: SNIP > > +static int tramp_mremap(const struct vm_special_mapping *sm, struct vm_area_struct *new_vma) > > +{ > > + return -EPERM; > > +} > > + > > +static struct vm_special_mapping tramp_mapping = { > > + .name = "[uprobes-trampoline]", > > + .mremap = tramp_mremap, > > +}; > > + > > +SYSCALL_DEFINE0(uprobe) > > +{ > > + struct pt_regs *regs = task_pt_regs(current); > > + struct vm_area_struct *vma; > > + unsigned long bp_vaddr; > > + int err; > > + > > + err = copy_from_user(&bp_vaddr, (void __user *)regs->sp + 3*8, sizeof(bp_vaddr)); > > A #define for the magic values would be nice. the 3*8 is to skip 3 values pushed on stack and get the return ip value, I'd prefer to keep 3*8 but it's definitely missing explaining comment above, wdyt? > > > + if (err) { > > + force_sig(SIGILL); > > + return -1; > > + } > > + > > + /* Allow execution only from uprobe trampolines. */ > > + vma = vma_lookup(current->mm, regs->ip); > > + if (!vma || vma->vm_private_data != (void *) &tramp_mapping) { > > vma_is_special_mapping() did not know about this function, thanks > > > + force_sig(SIGILL); > > + return -1; > > + } > > + > > + handle_syscall_uprobe(regs, bp_vaddr - 5); > > + return 0; > > +} > > + > > +asm ( > > + ".pushsection .rodata\n" > > + ".global uprobe_trampoline_entry\n" > > + "uprobe_trampoline_entry:\n" > > + "endbr64\n" > > + "push %rcx\n" > > + "push %r11\n" > > + "push %rax\n" > > + "movq $" __stringify(__NR_uprobe) ", %rax\n" > > + "syscall\n" > > + "pop %rax\n" > > + "pop %r11\n" > > + "pop %rcx\n" > > + "ret\n" > > + ".global uprobe_trampoline_end\n" > > + "uprobe_trampoline_end:\n" > > + ".popsection\n" > > +); > > + > > +extern __visible u8 uprobe_trampoline_entry[]; > > +extern __visible u8 uprobe_trampoline_end[]; > > + > > +const struct vm_special_mapping *arch_uprobe_trampoline_mapping(void) > > +{ > > + struct pt_regs *regs = task_pt_regs(current); > > + > > + return user_64bit_mode(regs) ? &tramp_mapping : NULL; > > +} > > + > > +static int __init arch_uprobes_init(void) > > +{ > > + unsigned long size = uprobe_trampoline_end - uprobe_trampoline_entry; > > + static struct page *pages[2]; > > + struct page *page; > > + > > + page = alloc_page(GFP_HIGHUSER); > > That page could be in static memory, removing the need for the explicit > allocation. It could also be __ro_after_init. > Then tramp_mapping itself can be const. hum, how would that look like? I think that to get proper page object you have to call alloc_page or some other page alloc family function.. what do I miss? > > Also this seems to waste the page on 32bit kernels. it's inside CONFIG_X86_64 ifdef > > > + if (!page) > > + return -ENOMEM; > > + pages[0] = page; > > + tramp_mapping.pages = (struct page **) &pages; > > tramp_mapping.pages = pages; ? I think the compiler will cry about *pages[2] vs **pages types mismatch, but I'll double check that thanks, jirka > > > + arch_uprobe_copy_ixol(page, 0, uprobe_trampoline_entry, size); > > + return 0; > > +} > > + > > +late_initcall(arch_uprobes_init); > > + > > /* > > * If arch_uprobe->insn doesn't use rip-relative addressing, return > > * immediately. Otherwise, rewrite the instruction so that it accesses > > [..]
On 2024-12-13 15:51:44+0100, Jiri Olsa wrote: > On Fri, Dec 13, 2024 at 02:48:00PM +0100, Thomas Weißschuh wrote: > > SNIP > > > > +static int tramp_mremap(const struct vm_special_mapping *sm, struct vm_area_struct *new_vma) > > > +{ > > > + return -EPERM; > > > +} > > > + > > > +static struct vm_special_mapping tramp_mapping = { > > > + .name = "[uprobes-trampoline]", > > > + .mremap = tramp_mremap, > > > +}; > > > + > > > +SYSCALL_DEFINE0(uprobe) > > > +{ > > > + struct pt_regs *regs = task_pt_regs(current); > > > + struct vm_area_struct *vma; > > > + unsigned long bp_vaddr; > > > + int err; > > > + > > > + err = copy_from_user(&bp_vaddr, (void __user *)regs->sp + 3*8, sizeof(bp_vaddr)); > > > > A #define for the magic values would be nice. > > the 3*8 is to skip 3 values pushed on stack and get the return ip value, > I'd prefer to keep 3*8 but it's definitely missing explaining comment > above, wdyt? A comment sounds good. > > > + if (err) { > > > + force_sig(SIGILL); > > > + return -1; > > > + } > > > + > > > + /* Allow execution only from uprobe trampolines. */ > > > + vma = vma_lookup(current->mm, regs->ip); > > > + if (!vma || vma->vm_private_data != (void *) &tramp_mapping) { > > > > vma_is_special_mapping() > > did not know about this function, thanks > > > > > > + force_sig(SIGILL); > > > + return -1; > > > + } > > > + > > > + handle_syscall_uprobe(regs, bp_vaddr - 5); > > > + return 0; > > > +} > > > + > > > +asm ( > > > + ".pushsection .rodata\n" > > > + ".global uprobe_trampoline_entry\n" > > > + "uprobe_trampoline_entry:\n" > > > + "endbr64\n" > > > + "push %rcx\n" > > > + "push %r11\n" > > > + "push %rax\n" > > > + "movq $" __stringify(__NR_uprobe) ", %rax\n" > > > + "syscall\n" > > > + "pop %rax\n" > > > + "pop %r11\n" > > > + "pop %rcx\n" > > > + "ret\n" > > > + ".global uprobe_trampoline_end\n" > > > + "uprobe_trampoline_end:\n" > > > + ".popsection\n" > > > +); > > > + > > > +extern __visible u8 uprobe_trampoline_entry[]; > > > +extern __visible u8 uprobe_trampoline_end[]; > > > + > > > +const struct vm_special_mapping *arch_uprobe_trampoline_mapping(void) > > > +{ > > > + struct pt_regs *regs = task_pt_regs(current); > > > + > > > + return user_64bit_mode(regs) ? &tramp_mapping : NULL; > > > +} > > > + > > > +static int __init arch_uprobes_init(void) > > > +{ > > > + unsigned long size = uprobe_trampoline_end - uprobe_trampoline_entry; > > > + static struct page *pages[2]; > > > + struct page *page; > > > + > > > + page = alloc_page(GFP_HIGHUSER); > > > > That page could be in static memory, removing the need for the explicit > > allocation. It could also be __ro_after_init. > > Then tramp_mapping itself can be const. > > hum, how would that look like? I think that to get proper page object > you have to call alloc_page or some other page alloc family function.. > what do I miss? static u8 trampoline_page[PAGE_SIZE] __ro_after_init __aligned(PAGE_SIZE); static struct page *tramp_mapping_pages[2] __ro_after_init; static const struct vm_special_mapping tramp_mapping = { .name = "[uprobes-trampoline]", .pages = tramp_mapping_pages, .mremap = tramp_mremap, }; static int __init arch_uprobes_init(void) { ... trampoline_pages[0] = virt_to_page(trampoline_page); ... } Untested, but it's similar to the stuff the vDSO implementations are doing which I am working with at the moment. > > > > Also this seems to waste the page on 32bit kernels. > > it's inside CONFIG_X86_64 ifdef > > > > > > + if (!page) > > > + return -ENOMEM; > > > + pages[0] = page; > > > + tramp_mapping.pages = (struct page **) &pages; > > > > tramp_mapping.pages = pages; ? > > I think the compiler will cry about *pages[2] vs **pages types mismatch, > but I'll double check that It compiles for me. > thanks, > jirka > > > > > > + arch_uprobe_copy_ixol(page, 0, uprobe_trampoline_entry, size); > > > + return 0; > > > +} > > > + > > > +late_initcall(arch_uprobes_init); > > > + > > > /* > > > * If arch_uprobe->insn doesn't use rip-relative addressing, return > > > * immediately. Otherwise, rewrite the instruction so that it accesses > > > > [..]
On Fri, Dec 13, 2024 at 04:12:46PM +0100, Thomas Weißschuh wrote: SNIP > > > > +static int __init arch_uprobes_init(void) > > > > +{ > > > > + unsigned long size = uprobe_trampoline_end - uprobe_trampoline_entry; > > > > + static struct page *pages[2]; > > > > + struct page *page; > > > > + > > > > + page = alloc_page(GFP_HIGHUSER); > > > > > > That page could be in static memory, removing the need for the explicit > > > allocation. It could also be __ro_after_init. > > > Then tramp_mapping itself can be const. > > > > hum, how would that look like? I think that to get proper page object > > you have to call alloc_page or some other page alloc family function.. > > what do I miss? > > static u8 trampoline_page[PAGE_SIZE] __ro_after_init __aligned(PAGE_SIZE); > static struct page *tramp_mapping_pages[2] __ro_after_init; > > static const struct vm_special_mapping tramp_mapping = { > .name = "[uprobes-trampoline]", > .pages = tramp_mapping_pages, > .mremap = tramp_mremap, > }; > > static int __init arch_uprobes_init(void) > { > ... > trampoline_pages[0] = virt_to_page(trampoline_page); > ... > } > > Untested, but it's similar to the stuff the vDSO implementations are > doing which I am working with at the moment. nice idea, better than allocating the page, will do that > > > > > > > Also this seems to waste the page on 32bit kernels. > > > > it's inside CONFIG_X86_64 ifdef > > > > > > > > > + if (!page) > > > > + return -ENOMEM; > > > > + pages[0] = page; > > > > + tramp_mapping.pages = (struct page **) &pages; > > > > > > tramp_mapping.pages = pages; ? > > > > I think the compiler will cry about *pages[2] vs **pages types mismatch, > > but I'll double check that > > It compiles for me. ok thanks, jirka
On 2024-12-13 22:52:15+0100, Jiri Olsa wrote: > On Fri, Dec 13, 2024 at 04:12:46PM +0100, Thomas Weißschuh wrote: > > SNIP > > > > > > +static int __init arch_uprobes_init(void) > > > > > +{ > > > > > + unsigned long size = uprobe_trampoline_end - uprobe_trampoline_entry; > > > > > + static struct page *pages[2]; > > > > > + struct page *page; > > > > > + > > > > > + page = alloc_page(GFP_HIGHUSER); > > > > > > > > That page could be in static memory, removing the need for the explicit > > > > allocation. It could also be __ro_after_init. > > > > Then tramp_mapping itself can be const. > > > > > > hum, how would that look like? I think that to get proper page object > > > you have to call alloc_page or some other page alloc family function.. > > > what do I miss? > > > > static u8 trampoline_page[PAGE_SIZE] __ro_after_init __aligned(PAGE_SIZE); > > static struct page *tramp_mapping_pages[2] __ro_after_init; > > > > static const struct vm_special_mapping tramp_mapping = { > > .name = "[uprobes-trampoline]", > > .pages = tramp_mapping_pages, > > .mremap = tramp_mremap, > > }; > > > > static int __init arch_uprobes_init(void) > > { > > ... > > trampoline_pages[0] = virt_to_page(trampoline_page); > > ... > > } > > > > Untested, but it's similar to the stuff the vDSO implementations are > > doing which I am working with at the moment. > > nice idea, better than allocating the page, will do that Or even better yet, just allocate the whole page already in the inline asm and avoid the copying, too: diff --git a/arch/x86/kernel/uprobes.c b/arch/x86/kernel/uprobes.c index b2420eeee23a..c5e6ca7f998a 100644 --- a/arch/x86/kernel/uprobes.c +++ b/arch/x86/kernel/uprobes.c @@ -462,7 +462,7 @@ SYSCALL_DEFINE0(uprobe) asm ( ".pushsection .rodata\n" - ".global uprobe_trampoline_entry\n" + ".balign " __stringify(PAGE_SIZE) "\n" "uprobe_trampoline_entry:\n" "endbr64\n" "push %rcx\n" @@ -474,13 +474,11 @@ asm ( "pop %r11\n" "pop %rcx\n" "ret\n" - ".global uprobe_trampoline_end\n" - "uprobe_trampoline_end:\n" + ".balign " __stringify(PAGE_SIZE) "\n" ".popsection\n" ); -extern __visible u8 uprobe_trampoline_entry[]; -extern __visible u8 uprobe_trampoline_end[]; +extern u8 uprobe_trampoline_entry[]; If you want to keep the copying for some reason, the asm code should be in the section ".init.rodata" as its not used afterwards. (A bit bikesheddy, I admit) Thomas
On Sat, Dec 14, 2024 at 02:21:43PM +0100, Thomas Weißschuh wrote: > On 2024-12-13 22:52:15+0100, Jiri Olsa wrote: > > On Fri, Dec 13, 2024 at 04:12:46PM +0100, Thomas Weißschuh wrote: > > > > SNIP > > > > > > > > +static int __init arch_uprobes_init(void) > > > > > > +{ > > > > > > + unsigned long size = uprobe_trampoline_end - uprobe_trampoline_entry; > > > > > > + static struct page *pages[2]; > > > > > > + struct page *page; > > > > > > + > > > > > > + page = alloc_page(GFP_HIGHUSER); > > > > > > > > > > That page could be in static memory, removing the need for the explicit > > > > > allocation. It could also be __ro_after_init. > > > > > Then tramp_mapping itself can be const. > > > > > > > > hum, how would that look like? I think that to get proper page object > > > > you have to call alloc_page or some other page alloc family function.. > > > > what do I miss? > > > > > > static u8 trampoline_page[PAGE_SIZE] __ro_after_init __aligned(PAGE_SIZE); > > > static struct page *tramp_mapping_pages[2] __ro_after_init; > > > > > > static const struct vm_special_mapping tramp_mapping = { > > > .name = "[uprobes-trampoline]", > > > .pages = tramp_mapping_pages, > > > .mremap = tramp_mremap, > > > }; > > > > > > static int __init arch_uprobes_init(void) > > > { > > > ... > > > trampoline_pages[0] = virt_to_page(trampoline_page); > > > ... > > > } > > > > > > Untested, but it's similar to the stuff the vDSO implementations are > > > doing which I am working with at the moment. > > > > nice idea, better than allocating the page, will do that > > Or even better yet, just allocate the whole page already in the inline > asm and avoid the copying, too: > > diff --git a/arch/x86/kernel/uprobes.c b/arch/x86/kernel/uprobes.c > index b2420eeee23a..c5e6ca7f998a 100644 > --- a/arch/x86/kernel/uprobes.c > +++ b/arch/x86/kernel/uprobes.c > @@ -462,7 +462,7 @@ SYSCALL_DEFINE0(uprobe) > > asm ( > ".pushsection .rodata\n" > - ".global uprobe_trampoline_entry\n" > + ".balign " __stringify(PAGE_SIZE) "\n" > "uprobe_trampoline_entry:\n" > "endbr64\n" > "push %rcx\n" > @@ -474,13 +474,11 @@ asm ( > "pop %r11\n" > "pop %rcx\n" > "ret\n" > - ".global uprobe_trampoline_end\n" > - "uprobe_trampoline_end:\n" > + ".balign " __stringify(PAGE_SIZE) "\n" > ".popsection\n" > ); > > -extern __visible u8 uprobe_trampoline_entry[]; > -extern __visible u8 uprobe_trampoline_end[]; > +extern u8 uprobe_trampoline_entry[]; > > > If you want to keep the copying for some reason, the asm code should be > in the section ".init.rodata" as its not used afterwards. perfect, no need for copy, I'll do what you propose above > > (A bit bikesheddy, I admit) thanks for the review, jirka
diff --git a/arch/x86/entry/syscalls/syscall_64.tbl b/arch/x86/entry/syscalls/syscall_64.tbl index 5eb708bff1c7..88e388c7675b 100644 --- a/arch/x86/entry/syscalls/syscall_64.tbl +++ b/arch/x86/entry/syscalls/syscall_64.tbl @@ -345,6 +345,7 @@ 333 common io_pgetevents sys_io_pgetevents 334 common rseq sys_rseq 335 common uretprobe sys_uretprobe +336 common uprobe sys_uprobe # don't use numbers 387 through 423, add new calls after the last # 'common' entry 424 common pidfd_send_signal sys_pidfd_send_signal diff --git a/arch/x86/kernel/uprobes.c b/arch/x86/kernel/uprobes.c index 22a17c149a55..23e4f2821cff 100644 --- a/arch/x86/kernel/uprobes.c +++ b/arch/x86/kernel/uprobes.c @@ -425,6 +425,86 @@ SYSCALL_DEFINE0(uretprobe) return -1; } +static int tramp_mremap(const struct vm_special_mapping *sm, struct vm_area_struct *new_vma) +{ + return -EPERM; +} + +static struct vm_special_mapping tramp_mapping = { + .name = "[uprobes-trampoline]", + .mremap = tramp_mremap, +}; + +SYSCALL_DEFINE0(uprobe) +{ + struct pt_regs *regs = task_pt_regs(current); + struct vm_area_struct *vma; + unsigned long bp_vaddr; + int err; + + err = copy_from_user(&bp_vaddr, (void __user *)regs->sp + 3*8, sizeof(bp_vaddr)); + if (err) { + force_sig(SIGILL); + return -1; + } + + /* Allow execution only from uprobe trampolines. */ + vma = vma_lookup(current->mm, regs->ip); + if (!vma || vma->vm_private_data != (void *) &tramp_mapping) { + force_sig(SIGILL); + return -1; + } + + handle_syscall_uprobe(regs, bp_vaddr - 5); + return 0; +} + +asm ( + ".pushsection .rodata\n" + ".global uprobe_trampoline_entry\n" + "uprobe_trampoline_entry:\n" + "endbr64\n" + "push %rcx\n" + "push %r11\n" + "push %rax\n" + "movq $" __stringify(__NR_uprobe) ", %rax\n" + "syscall\n" + "pop %rax\n" + "pop %r11\n" + "pop %rcx\n" + "ret\n" + ".global uprobe_trampoline_end\n" + "uprobe_trampoline_end:\n" + ".popsection\n" +); + +extern __visible u8 uprobe_trampoline_entry[]; +extern __visible u8 uprobe_trampoline_end[]; + +const struct vm_special_mapping *arch_uprobe_trampoline_mapping(void) +{ + struct pt_regs *regs = task_pt_regs(current); + + return user_64bit_mode(regs) ? &tramp_mapping : NULL; +} + +static int __init arch_uprobes_init(void) +{ + unsigned long size = uprobe_trampoline_end - uprobe_trampoline_entry; + static struct page *pages[2]; + struct page *page; + + page = alloc_page(GFP_HIGHUSER); + if (!page) + return -ENOMEM; + pages[0] = page; + tramp_mapping.pages = (struct page **) &pages; + arch_uprobe_copy_ixol(page, 0, uprobe_trampoline_entry, size); + return 0; +} + +late_initcall(arch_uprobes_init); + /* * If arch_uprobe->insn doesn't use rip-relative addressing, return * immediately. Otherwise, rewrite the instruction so that it accesses diff --git a/include/linux/syscalls.h b/include/linux/syscalls.h index c6333204d451..002f4e1debe5 100644 --- a/include/linux/syscalls.h +++ b/include/linux/syscalls.h @@ -994,6 +994,8 @@ asmlinkage long sys_ioperm(unsigned long from, unsigned long num, int on); asmlinkage long sys_uretprobe(void); +asmlinkage long sys_uprobe(void); + /* pciconfig: alpha, arm, arm64, ia64, sparc */ asmlinkage long sys_pciconfig_read(unsigned long bus, unsigned long dfn, unsigned long off, unsigned long len, diff --git a/include/linux/uprobes.h b/include/linux/uprobes.h index c4ee755ca2a1..5e9a33bfb747 100644 --- a/include/linux/uprobes.h +++ b/include/linux/uprobes.h @@ -232,6 +232,7 @@ extern struct uprobe_trampoline *uprobe_trampoline_get(unsigned long vaddr); extern void uprobe_trampoline_put(struct uprobe_trampoline *area); extern bool arch_uprobe_is_callable(unsigned long vtramp, unsigned long vaddr); extern const struct vm_special_mapping *arch_uprobe_trampoline_mapping(void); +extern void handle_syscall_uprobe(struct pt_regs *regs, unsigned long bp_vaddr); #else /* !CONFIG_UPROBES */ struct uprobes_state { }; diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c index f57918c624da..52f38d1ef276 100644 --- a/kernel/events/uprobes.c +++ b/kernel/events/uprobes.c @@ -2729,6 +2729,28 @@ static void handle_swbp(struct pt_regs *regs) rcu_read_unlock_trace(); } +void handle_syscall_uprobe(struct pt_regs *regs, unsigned long bp_vaddr) +{ + struct uprobe *uprobe; + int is_swbp; + + rcu_read_lock_trace(); + uprobe = find_active_uprobe_rcu(bp_vaddr, &is_swbp); + if (!uprobe) + goto unlock; + + if (!get_utask()) + goto unlock; + + if (arch_uprobe_ignore(&uprobe->arch, regs)) + goto unlock; + + handler_chain(uprobe, regs); + + unlock: + rcu_read_unlock_trace(); +} + /* * Perform required fix-ups and disable singlestep. * Allow pending signals to take effect. diff --git a/kernel/sys_ni.c b/kernel/sys_ni.c index c00a86931f8c..bf5d05c635ff 100644 --- a/kernel/sys_ni.c +++ b/kernel/sys_ni.c @@ -392,3 +392,4 @@ COND_SYSCALL(setuid16); COND_SYSCALL(rseq); COND_SYSCALL(uretprobe); +COND_SYSCALL(uprobe);
Adding new uprobe syscall that calls uprobe handlers for given 'breakpoint' address. The idea is that the 'breakpoint' address calls the user space trampoline which executes the uprobe syscall. The syscall handler reads the return address of the initial call to retrieve the original 'breakpoint' address. With this address we find the related uprobe object and call its consumers. Adding the arch_uprobe_trampoline_mapping function that provides uprobe trampoline mapping. This mapping is backed with one global page initialized at __init time and shared by the all the mapping instances. We do not allow to execute uprobe syscall if the caller is not from uprobe trampoline mapping. Signed-off-by: Jiri Olsa <jolsa@kernel.org> --- arch/x86/entry/syscalls/syscall_64.tbl | 1 + arch/x86/kernel/uprobes.c | 80 ++++++++++++++++++++++++++ include/linux/syscalls.h | 2 + include/linux/uprobes.h | 1 + kernel/events/uprobes.c | 22 +++++++ kernel/sys_ni.c | 1 + 6 files changed, 107 insertions(+)