From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ozlabs.org (bilbo.ozlabs.org [IPv6:2401:3900:2:1::2]) (using TLSv1.2 with cipher ADH-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by lists.ozlabs.org (Postfix) with ESMTPS id 41ptdT01RTzF0WK for ; Mon, 13 Aug 2018 21:23:25 +1000 (AEST) In-Reply-To: <20180810122535.11710-1-mpe@ellerman.id.au> To: Michael Ellerman , linuxppc-dev@ozlabs.org From: Michael Ellerman Subject: Re: powerpc/uaccess: Enable get_user(u64, *p) on 32-bit Message-Id: <41ptdS0svWz9sCR@ozlabs.org> Date: Mon, 13 Aug 2018 21:23:23 +1000 (AEST) List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Fri, 2018-08-10 at 12:25:35 UTC, Michael Ellerman wrote: > Currently if you build a 32-bit powerpc kernel and use get_user() to > load a u64 value it will fail to build with eg: > > kernel/rseq.o: In function `rseq_get_rseq_cs': > kernel/rseq.c:123: undefined reference to `__get_user_bad' > > This is hitting the check in __get_user_size() that makes sure the > size we're copying doesn't exceed the size of the destination: > > #define __get_user_size(x, ptr, size, retval) > do { > retval = 0; > __chk_user_ptr(ptr); > if (size > sizeof(x)) > (x) = __get_user_bad(); > > Which doesn't immediately make sense because the size of the > destination is u64, but it's not really, because __get_user_check() > etc. internally create an unsigned long and copy into that: > > #define __get_user_check(x, ptr, size) > ({ > long __gu_err = -EFAULT; > unsigned long __gu_val = 0; > > The problem being that on 32-bit unsigned long is not big enough to > hold a u64. We can fix this with a trick from hpa in the x86 code, we > statically check the type of x and set the type of __gu_val to either > unsigned long or unsigned long long. > > Signed-off-by: Michael Ellerman Applied to powerpc next. https://git.kernel.org/powerpc/c/f7a6947cd49b7ff4e03f1b4f7e7b22 cheers