[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20160428191914.GA28527@yury-N73SV>
Date: Thu, 28 Apr 2016 22:19:14 +0300
From: Yury Norov <ynorov@...iumnetworks.com>
To: Catalin Marinas <catalin.marinas@....com>
CC: <arnd@...db.de>, <linux-arm-kernel@...ts.infradead.org>,
<linux-kernel@...r.kernel.org>, <linux-arch@...r.kernel.org>,
<linux-s390@...r.kernel.org>, <pinskia@...il.com>,
<Prasun.Kapoor@...iumnetworks.com>, <schwab@...e.de>,
<broonie@...nel.org>, <linux-doc@...r.kernel.org>,
<heiko.carstens@...ibm.com>, <agraf@...e.de>,
<klimov.linux@...il.com>,
Andrew Pinski <Andrew.Pinski@...iumnetworks.com>,
<bamvor.zhangjian@...wei.com>, Andrew Pinski <apinski@...ium.com>,
<schwidefsky@...ibm.com>, <Nathan_Lynch@...tor.com>,
<joseph@...esourcery.com>,
<christoph.muellner@...obroma-systems.com>
Subject: Re: [PATCH 20/25] arm64:ilp32: add sys_ilp32.c and a separate table
(in entry.S) to use it
On Tue, Apr 26, 2016 at 05:57:01PM +0100, Catalin Marinas wrote:
> On Wed, Apr 06, 2016 at 01:08:42AM +0300, Yury Norov wrote:
> > +/* Using non-compat syscalls where necessary */
> > +#define compat_sys_fadvise64_64 sys_fadvise64_64
> > +#define compat_sys_fallocate sys_fallocate
> > +#define compat_sys_ftruncate64 sys_ftruncate
> > +#define compat_sys_lookup_dcookie sys_lookup_dcookie
> > +#define compat_sys_pread64 sys_pread64
> > +#define compat_sys_pwrite64 sys_pwrite64
> > +#define compat_sys_readahead sys_readahead
> > +#define compat_sys_shmat sys_shmat
>
> Why don't we use compat_sys_shmat? Is it because of COMPAT_SHMLBA?
Yes. COMPAT_SHMLBA is 4 pages, and it's aarch32-only limitation.
>
> > +#define compat_sys_sync_file_range sys_sync_file_range
> > +#define compat_sys_truncate64 sys_truncate
> > +#define sys_llseek sys_lseek
> > +#define sys_mmap2 sys_mmap
>
> Nitpick: there are some whitespace inconsistencies above (just convert
> all spaces to tabs).
>
> I think you should also update Documentation/arm64/ilp32.txt to include
> the list above.
OK
>
> > +
> > +#include <asm/syscall.h>
> > +
> > +#undef __SYSCALL
> > +#undef __SC_COMP
> > +#undef __SC_WRAP
> > +#undef __SC_3264
> > +#undef __SC_COMP_3264
>
> Minor detail: do we actually need to undef all these? Maybe we can get
> away with just defining __SYSCALL_COMPAT at the top of the file.
>
Yes, we need. Otherwise we have circular dependency like this:
arch/arm64/kernel/sys_ilp32.c:60:0: warning: "__SC_WRAP" redefined
#define __SC_WRAP(nr, sym) [nr] = compat_##sym,
^
In file included from include/asm-generic/unistd.h:1:0,
from ./arch/arm64/include/uapi/asm/unistd.h:16,
from ./arch/arm64/include/asm/unistd.h:62,
from ./include/uapi/linux/unistd.h:7,
from include/linux/syscalls.h:23,
from arch/arm64/kernel/sys_ilp32.c:30:
include/uapi/asm-generic/unistd.h:33:0: note: this is the location of the previous definition
#define __SC_WRAP __SYSCALL
Defining __SYSCALL_COMPAT at the top of the file does not help much.
> > +
> > +#define __SYSCALL_COMPAT
> > +#define __SYSCALL(nr, sym) [nr] = sym,
> > +#define __SC_WRAP(nr, sym) [nr] = compat_##sym,
> > +
> > +/*
> > + * The sys_call_ilp32_table array must be 4K aligned to be accessible from
> > + * kernel/entry.S.
> > + */
> > +void *sys_call_ilp32_table[__NR_syscalls] __aligned(4096) = {
> > + [0 ... __NR_syscalls - 1] = sys_ni_syscall,
> > +#include <asm/unistd.h>
> > +};
>
> --
> Catalin
Powered by blists - more mailing lists