riscv: fix seccomp reject syscall code path
Subsystems (2)
PATCH 1/1 riscv: fix seccomp reject syscall code path
If secure_computing() rejected a system call, we were previously setting the system call number to -1, to indicate to later code that the syscall failed. However, if something (e.g. a user notification) was sleeping, and received a signal, we may set a0 to -ERESTARTSYS and re-try the system call again.
In this case, seccomp "denies" the syscall (because of the signal), and we would set a7 to -1, thus losing the value of the system call we want to restart.
Instead, let's return -1 from do_syscall_trace_enter() to indicate that the syscall was rejected, so we don't clobber the value in case of -ERESTARTSYS or whatever.
This commit fixes the user_notification_signal seccomp selftest on riscv to no longer hang. That test expects the system call to be re-issued after the signal, and it wasn't due to the above bug. Now that it is, everything works normally.
Note that in the ptrace (tracer) case, the tracer can set the register values to whatever they want, so we still need to keep the code that handles out-of-bounds syscalls. However, we can drop the comment.
We can also drop syscall_set_nr(), since it is no longer used anywhere, and the code that re-loads the value in a7 because of it.
Reported in: https://lore.kernel.org/bpf/CAEn-LTp=ss0Dfv6J00=rCAy+N78U2AmhqJNjfqjr2FDpPYjxEQ@mail.gmail.com/
--- arch/riscv/include/asm/syscall.h | 7 ------- arch/riscv/kernel/entry.S | 11 +++-------- arch/riscv/kernel/ptrace.c | 11 +++++------ 3 files changed, 8 insertions(+), 21 deletions(-)
arch/riscv/include/asm/syscall.h +0 -7
@@ -28,13 +28,6 @@ static inline int syscall_get_nr(struct task_struct *task,return regs->a7; } -static inline void syscall_set_nr (struct task_struct *task, -struct pt_regs *regs, -int sysno) -{ - regs->a7 = sysno; -} -static inline void syscall_rollback (struct task_struct *task,struct pt_regs *regs) {
arch/riscv/kernel/entry.S +3 -8
@@ -228,20 +228,13 @@ check_syscall_nr: /* Check to make sure we don't jump to a bogus syscall number. */ li t0, __NR_syscalls la s0, sys_ni_syscall - /* - * The tracer can change syscall number to valid/invalid value. - * We use syscall_set_nr helper in syscall_trace_enter thus we - * cannot trust the current value in a7 and have to reload from - * the current task pt_regs. - */ - REG_L a7, PT_A7(sp) /* * Syscall number held in a7. * If syscall number is above allowed value, redirect to ni_syscall. */ bge a7, t0, 1f /* - * Check if syscall is rejected by tracer or seccomp, i.e., a7 == -1. + * Check if syscall is rejected by tracer, i.e., a7 == -1. * If yes, we pretend it was executed. */ li t1, -1
@@ -334,6 +327,7 @@ work_resched: handle_syscall_trace_enter: move a0, sp call do_syscall_trace_enter + move t0, a0 REG_L a0, PT_A0(sp) REG_L a1, PT_A1(sp) REG_L a2, PT_A2(sp)
@@ -342,6 +336,7 @@ handle_syscall_trace_enter: REG_L a5, PT_A5(sp) REG_L a6, PT_A6(sp) REG_L a7, PT_A7(sp) + bnez t0, ret_from_syscall_rejected j check_syscall_nr handle_syscall_trace_exit: move a0, sp
arch/riscv/kernel/ptrace.c +5 -6
@@ -148,21 +148,19 @@ long arch_ptrace(struct task_struct *child, long request,* Allows PTRACE_SYSCALL to work. These are called from entry.S in * {handle,ret_from}_syscall. */ -__visible void do_syscall_trace_enter(struct pt_regs *regs) +__visible int do_syscall_trace_enter(struct pt_regs *regs) {if (test_thread_flag (TIF_SYSCALL_TRACE ))if (tracehook_report_syscall_entry (regs)) -syscall_set_nr (current, regs,-1 ); +return -1 ;/* * Do the secure computing after ptrace; failures should be fast. * If this fails we might have return value in a0 from seccomp * (via SECCOMP_RET_ERRNO/TRACE). */ -if (secure_computing () ==-1 ) { -syscall_set_nr (current, regs,-1 ); -return ; - } +if (secure_computing () ==-1 ) +return -1 ;#ifdef CONFIG_HAVE_SYSCALL_TRACEPOINTS if (test_thread_flag (TIF_SYSCALL_TRACEPOINT ))
@@ -170,6 +168,7 @@ __visible void do_syscall_trace_enter(struct pt_regs *regs)#endif audit_syscall_entry (regs->a7, regs->a0, regs->a1, regs->a2, regs->a3); +return 0 ; } __visible void do_syscall_trace_exit(struct pt_regs *regs)
Signature · 1 line
2.20.1
Quoted 23 lines · Tycho Andersen
If
secure_computing()rejected a system call, we were previously setting the system call number to -1, to indicate to later code that the syscall failed. However, if something (e.g. a user notification) was sleeping, and received a signal, we may set a0 to-ERESTARTSYSand re-try the system call again.In this case, seccomp "denies" the syscall (because of the signal), and we would set a7 to -1, thus losing the value of the system call we want to restart.
Instead, let's return -1 from
do_syscall_trace_enter()to indicate that the syscall was rejected, so we don't clobber the value in case of-ERESTARTSYSor whatever.This commit fixes the
user_notification_signalseccomp selftest on riscv to no longer hang. That test expects the system call to be re-issued after the signal, and it wasn't due to the above bug. Now that it is, everything works normally.Note that in the ptrace (tracer) case, the tracer can set the register values to whatever they want, so we still need to keep the code that handles out-of-bounds syscalls. However, we can drop the comment.
Quoted 7 lines
We can also drop
syscall_set_nr(), since it is no longer used anywhere, and the code that re-loads the value in a7 because of it.Reported in: https://lore.kernel.org/bpf/CAEn-LTp=ss0Dfv6J00=rCAy+N78U2AmhqJNjfqjr2FDpPYjxEQ@mail.gmail.com/
Reported-by David Abdurachmanov <[email protected]>Signed-off-by Tycho Andersen <[email protected]>
Funky! Good catch. :)
-Kees
Quoted 111 lines
CC Kees Cook <[email protected]>CC Andy Lutomirski <[email protected]>CC Paul Walmsley <[email protected]>CC Oleg Nesterov <[email protected]>--- arch/riscv/include/asm/syscall.h | 7 ------- arch/riscv/kernel/entry.S | 11 +++-------- arch/riscv/kernel/ptrace.c | 11 +++++------ 3 files changed, 8 insertions(+), 21 deletions(-)arch/riscv/include/asm/syscall.h +0
-7@@ -28,13 +28,6 @@ static inline int syscall_get_nr(struct task_struct *task,return regs->a7; } -static inline void syscall_set_nr (struct task_struct *task, -struct pt_regs *regs, -int sysno) -{ - regs->a7 = sysno; -} -static inline void syscall_rollback (struct task_struct *task,struct pt_regs *regs) {arch/riscv/kernel/entry.S +3
-8@@ -228,20 +228,13 @@ check_syscall_nr: /* Check to make sure we don't jump to a bogus syscall number. */ li t0, __NR_syscalls la s0, sys_ni_syscall - /* - * The tracer can change syscall number to valid/invalid value. - * We use syscall_set_nr helper in syscall_trace_enter thus we - * cannot trust the current value in a7 and have to reload from - * the current task pt_regs. - */ - REG_L a7, PT_A7(sp) /* * Syscall number held in a7. * If syscall number is above allowed value, redirect to ni_syscall. */ bge a7, t0, 1f /* - * Check if syscall is rejected by tracer or seccomp, i.e., a7 == -1. + * Check if syscall is rejected by tracer, i.e., a7 == -1. * If yes, we pretend it was executed. */ li t1, -1@@ -334,6 +327,7 @@ work_resched: handle_syscall_trace_enter: move a0, sp call do_syscall_trace_enter + move t0, a0 REG_L a0, PT_A0(sp) REG_L a1, PT_A1(sp) REG_L a2, PT_A2(sp)@@ -342,6 +336,7 @@ handle_syscall_trace_enter: REG_L a5, PT_A5(sp) REG_L a6, PT_A6(sp) REG_L a7, PT_A7(sp) + bnez t0, ret_from_syscall_rejected j check_syscall_nr handle_syscall_trace_exit: move a0, sparch/riscv/kernel/ptrace.c +5
-6@@ -148,21 +148,19 @@ long arch_ptrace(struct task_struct *child, long request,* Allows PTRACE_SYSCALL to work. These are called from entry.S in * {handle,ret_from}_syscall. */ -__visible void do_syscall_trace_enter(struct pt_regs *regs) +__visible int do_syscall_trace_enter(struct pt_regs *regs) {if (test_thread_flag (TIF_SYSCALL_TRACE ))if (tracehook_report_syscall_entry (regs)) -syscall_set_nr (current, regs,-1 ); +return -1 ;/* * Do the secure computing after ptrace; failures should be fast. * If this fails we might have return value in a0 from seccomp * (via SECCOMP_RET_ERRNO/TRACE). */ -if (secure_computing () ==-1 ) { -syscall_set_nr (current, regs,-1 ); -return ; - } +if (secure_computing () ==-1 ) +return -1 ;#ifdef CONFIG_HAVE_SYSCALL_TRACEPOINTS if (test_thread_flag (TIF_SYSCALL_TRACEPOINT ))@@ -170,6 +168,7 @@ __visible void do_syscall_trace_enter(struct pt_regs *regs)#endif audit_syscall_entry (regs->a7, regs->a0, regs->a1, regs->a2, regs->a3); +return 0 ; } __visible void do_syscall_trace_exit(struct pt_regs *regs)Signature · 1 line
2.20.1
Signature · 1 line
Kees Cook
Quoted 1 line · Tycho Andersen
...
Ping, any risc-v people have thoughts on this?
Tycho
Quoted 6 lines · Tycho Andersen
Quoted 1 line · Tycho Andersen
...
Ping, any risc-v people have thoughts on this?
Tycho
Re-ping. :) Can someone please pick this up? Original patch here:
https://lore.kernel.org/lkml/[email protected]/
Signature · 1 line
Kees Cook
Palmer Dabbelt
maintainerQuoted 10 lines · On Mon, 02 Mar 2020 20:46:46 PST (-0800), [email protected] wrote:
Quoted 6 lines · Tycho Andersen
Quoted 1 line · Tycho Andersen
...
Ping, any risc-v people have thoughts on this?
Tycho
Re-ping. :) Can someone please pick this up? Original patch here:
https://lore.kernel.org/lkml/[email protected]/
Sorry, the other messages didn't end up in my inbox. I'll take a look, as this seems like a good candidate for rc5.
Quoted 7 lines · Palmer Dabbelt
Quoted 6 lines · On Mon, 02 Mar 2020 20:46:46 PST (-0800), [email protected] wrote:
Quoted 5 lines · Tycho Andersen
Quoted 1 line · Tycho Andersen
...
Ping, any risc-v people have thoughts on this?
Quoted 7 lines
Quoted 4 lines
Quoted 1 line
Tycho
Re-ping. :) Can someone please pick this up? Original patch here:
https://lore.kernel.org/lkml/[email protected]/Sorry, the other messages didn't end up in my inbox. I'll take a look, as this seems like a good candidate for rc5.
Awesome; thanks!
-Kees
Signature · 1 line
Kees Cook
Palmer Dabbelt
maintainerQuoted 4 lines · On Sun, 23 Feb 2020 09:17:57 PST (-0800), [email protected] wrote:
Quoted 1 line · Tycho Andersen
...
Ping, any risc-v people have thoughts on this?
Thanks for fixing this. It ended up escaping my inbox, but it's on fixes now and targeted for the next RC.