GNU linker's -z common-page-size's default value is based on the target
architecture. arch/sparc/vdso/Makefile sets it to the architecture
default, which is implicit and redundant. Drop it.

Link: https://lkml.kernel.org/r/20181206191231.192355-1-ndesaulniers@google.com
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Older versions of bintutils do not allow symbol math across different
segments on sparc:

====================
Assembler messages:
99: Error: operation combines symbols in different segments
====================

This is controlled by whether or not DIFF_EXPR_OK is defined in
gas/config/tc-*.h and for sparc this was not the case until mid-2017.

So we have to patch between %stick and %tick another way.

Do what powerpc does and emit two versions of the relevant functions,
one using %tick and one using %stick, and patch the symbols in the
dynamic symbol table.

Fixes: 2f6c9bf31a0b ("sparc: Improve VDSO instruction patching.")
Reported-by: Meelis Roos <mroos@linux.ee>
Tested-by: Meelis Roos <mroos@linux.ee>
Signed-off-by: David S. Miller <davem@davemloft.net>

Almost entirely borrowed from the x86 code.

Main improvement is to avoid having to initialize
ts->tv_nsec to zero before the sequence loops, by
expanding timespec_add_ns().

Signed-off-by: David S. Miller <davem@davemloft.net>

There should be no undefined symbols in the resulting VDSO image(s).

On sparc, fixed register usage can result in undefined symbols ending
up in the image.  To combat this, we do two things:

1) Define current_thread_info() specially when BUILD_DSO.

2) Ignore "#scratch" register undefined symbols in the output.

Signed-off-by: David S. Miller <davem@davemloft.net>

Rather than funneling through CC.

Also, use --hash-style=both just like other VDSO architectures and
glibc do.

Signed-off-by: David S. Miller <davem@davemloft.net>

Do not set any special register usage options, use the default which
is exactly what we should use for userspace code.

Make sure we remove the gcc plugin options from the 64-bit build.
The 32-bit cflags got it right already.

Signed-off-by: David S. Miller <davem@davemloft.net>

Not in vclock_gettime.c itself.

Signed-off-by: David S. Miller <davem@davemloft.net>

If the TICK_PRIV_BIT was set, we would not be able to read the tick
register in user space, which is where this code runs.

Signed-off-by: David S. Miller <davem@davemloft.net>

One interesting thing we need to do is stop using
__builtin_return_address() in get_vvar_data().

Simply read the %pc register instead.

Signed-off-by: David S. Miller <davem@davemloft.net>

The current VDSO patch mechanism has several problems:

1) It assumes how gcc will emit a function, with a register
   window, an initial save instruction and then immediately
   the %tick read when compiling vread_tick().

   There is no such guarantees, code generation could change
   at any time, gcc could put a nop between the save and
   the %tick read, etc.

   So this is extremely fragile and would fail some day.

2) It disallows us to properly inline vread_tick() into the callers
   and thus get the best possible code sequences.

So fix this to patch properly, with location based annotations.

We have to be careful because we cannot do it the way we do
patches elsewhere in the kernel.  Those use a sequence like:

	1:
	insn
	.section	.whatever_patch, "ax"
	.word		1b
	replacement_insn
	.previous

This is a dynamic shared object, so that .word cannot be resolved at
build time, and thus cannot be used to execute the patches when the
kernel initializes the images.

Even trying to use label difference equations doesn't work in the
above kind of scheme:

	1:
	insn
	.section	.whatever_patch, "ax"
	.word		. - 1b
	replacement_insn
	.previous

The assembler complains that it cannot resolve that computation.
The issue is that this is contained in an executable section.

Borrow the sequence used by x86 alternatives, which is:

	1:
	insn
	.pushsection	.whatever_patch, "a"
	.word		. - 1b, . - 1f
	.popsection
	.pushsection	.whatever_patch_replacements, "ax"
	1:
	replacement_insn
	.previous

This works, allows us to inline vread_tick() as much as we like, and
can be used for arbitrary kinds of VDSO patching in the future.

Also, reverse the condition for patching.  Most systems are %stick
based, so if we only patch on %tick systems the patching code will
get little or no testing.

Signed-off-by: David S. Miller <davem@davemloft.net>