[PATCH v7 8/9] ARM: vdso initialization, mapping, and synchronization

On Tue, Jul 1, 2014 at 7:15 AM, Will Deacon [off-list ref] wrote:

On Tue, Jul 01, 2014 at 03:11:04PM +0100, Nathan Lynch wrote:

quoted

On 07/01/2014 04:03 AM, Will Deacon wrote:

quoted

On Mon, Jun 30, 2014 at 10:37:48PM +0100, Andy Lutomirski wrote:

quoted

On 06/22/2014 08:11 PM, Nathan Lynch wrote:

quoted

Initialize the vdso page list at boot, install the vdso mapping at
exec time, and update the data page during timer ticks.  This code is
not built if CONFIG_VDSO is not enabled.

Account for the vdso length when randomizing the offset from the
stack.  The vdso is placed immediately following the sigpage with a
separate install_special_mapping call in arm_install_vdso.

[...]

quoted

quoted

+/* assumes mmap_sem is write-locked */
+void arm_install_vdso(struct mm_struct *mm, unsigned long addr)
+{
+ int ret;
+
+ mm->context.vdso = ~0UL;
+
+ if (vdso_pagelist == NULL)
+         return;
+
+ /*
+  * Put vDSO base into mm struct before calling
+  * install_special_mapping so the perf counter mmap tracking
+  * code will recognise it as a vDSO.
+  */
+ mm->context.vdso = addr;
+
+ ret = install_special_mapping(mm, addr, vdso_mapping_len,
+                               VM_READ|VM_EXEC|
+                               VM_MAYREAD|VM_MAYWRITE|VM_MAYEXEC,
+                               vdso_pagelist);

Eek.  You're mapping the shared data VM_MAYWRITE.  This will cause
bizarre and confusing failures if ptrace pokes at it.

Hmm, but how else can we support software breakpoints on the vdso?

I believe Andy is suggesting separate VMAs (with different VM flags) for
the VDSO's data and code.  So, breakpoints in code would work, but
attempts to modify the data page via ptrace() would fail outright
instead of silently COWing.

Ah, yes. That makes a lot of sense for the data page -- we should do
something similar on arm64 too, since the CoW will break everything for the
task being debugged. We could also drop the EXEC flags too.

If you do this, I have a slight preference for the new vma being
called "[vvar]" to match x86.  It'll make the CRIU people happy if and
when they port it to ARM.

--Andy