[PATCH V2] riscv: compat: Using seperated vdso_maps for compat_vdso_info

[Jessica Clarke]

On 26 May 2022, at 01:33, guoren at kernel.org wrote:
> 
> From: Guo Ren <guoren at linux.alibaba.com>
> 
> This is a fixup for vdso implementation which caused musl to
> fail.
> 
> [   11.600082] Run /sbin/init as init process
> [   11.628561] init[1]: unhandled signal 11 code 0x1 at
> 0x0000000000000000 in libc.so[ffffff8ad39000+a4000]
> [   11.629398] CPU: 0 PID: 1 Comm: init Not tainted
> 5.18.0-rc7-next-20220520 #1
> [   11.629462] Hardware name: riscv-virtio,qemu (DT)
> [   11.629546] epc : 00ffffff8ada1100 ra : 00ffffff8ada13c8 sp :
> 00ffffffc58199f0
> [   11.629586]  gp : 00ffffff8ad39000 tp : 00ffffff8ade0998 t0 :
> ffffffffffffffff
> [   11.629598]  t1 : 00ffffffc5819fd0 t2 : 0000000000000000 s0 :
> 00ffffff8ade0cc0
> [   11.629610]  s1 : 00ffffff8ade0cc0 a0 : 0000000000000000 a1 :
> 00ffffffc5819a00
> [   11.629622]  a2 : 0000000000000001 a3 : 000000000000001e a4 :
> 00ffffffc5819b00
> [   11.629634]  a5 : 00ffffffc5819b00 a6 : 0000000000000000 a7 :
> 0000000000000000
> [   11.629645]  s2 : 00ffffff8ade0ac8 s3 : 00ffffff8ade0ec8 s4 :
> 00ffffff8ade0728
> [   11.629656]  s5 : 00ffffff8ade0a90 s6 : 0000000000000000 s7 :
> 00ffffffc5819e40
> [   11.629667]  s8 : 00ffffff8ade0ca0 s9 : 00ffffff8addba50 s10:
> 0000000000000000
> [   11.629678]  s11: 0000000000000000 t3 : 0000000000000002 t4 :
> 0000000000000001
> [   11.629688]  t5 : 0000000000020000 t6 : ffffffffffffffff
> [   11.629699] status: 0000000000004020 badaddr: 0000000000000000
> cause: 000000000000000d
> 
> The last __vdso_init(&compat_vdso_info) replaces the data in normal
> vdso_info. This is an obvious bug.
> 
> Reported-by: Guenter Roeck <linux at roeck-us.net>
> Tested-by: Guenter Roeck <linux at roeck-us.net>
> Tested-by: Heiko Stübner <heiko at sntech.de>
> Signed-off-by: Guo Ren <guoren at linux.alibaba.com>
> Signed-off-by: Guo Ren <guoren at kernel.org>
> Cc: Palmer Dabbelt <palmer at dabbelt.com>
> ---
> Changes in V2:
> - Add Tested-by
> - Rename vvar & vdso in /proc/<pid>/maps.

Why? No other architecture renames it to that, and various pieces of
software look for the magic [vdso] name, including GDB and LLDB, though
by my count there are 57 source packages in Debian that contain the
string literal "[vdso]" (including quotes), including Firefox,
Android's ART, lvm2 and elfutils.

Jess

> ---
> arch/riscv/kernel/vdso.c | 15 +++++++++++++--
> 1 file changed, 13 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/riscv/kernel/vdso.c b/arch/riscv/kernel/vdso.c
> index 50fe4c877603..957f164c9778 100644
> --- a/arch/riscv/kernel/vdso.c
> +++ b/arch/riscv/kernel/vdso.c
> @@ -206,12 +206,23 @@ static struct __vdso_info vdso_info __ro_after_init = {
> };
> 
> #ifdef CONFIG_COMPAT
> +static struct vm_special_mapping rv_compat_vdso_maps[] __ro_after_init = {
> +	[RV_VDSO_MAP_VVAR] = {
> +		.name   = "[compat vvar]",
> +		.fault = vvar_fault,
> +	},
> +	[RV_VDSO_MAP_VDSO] = {
> +		.name   = "[compat vdso]",
> +		.mremap = vdso_mremap,
> +	},
> +};
> +
> static struct __vdso_info compat_vdso_info __ro_after_init = {
> 	.name = "compat_vdso",
> 	.vdso_code_start = compat_vdso_start,
> 	.vdso_code_end = compat_vdso_end,
> -	.dm = &rv_vdso_maps[RV_VDSO_MAP_VVAR],
> -	.cm = &rv_vdso_maps[RV_VDSO_MAP_VDSO],
> +	.dm = &rv_compat_vdso_maps[RV_VDSO_MAP_VVAR],
> +	.cm = &rv_compat_vdso_maps[RV_VDSO_MAP_VDSO],
> };
> #endif
> 
> -- 
> 2.36.1
> 
> 
> _______________________________________________
> linux-riscv mailing list
> linux-riscv at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-riscv