linux-kernel - Re: [PATCH] s390: fix build error for sys_call_table

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20200618115201.GA5671@willie-the-truck>
Date:   Thu, 18 Jun 2020 12:52:02 +0100
From:   Will Deacon <will@...nel.org>
To:     Heiko Carstens <heiko.carstens@...ibm.com>
Cc:     Xiaoming Ni <nixiaoming@...wei.com>, acme@...nel.org,
        alexander.shishkin@...ux.intel.com, arnd@...db.de,
        borntraeger@...ibm.com, catalin.marinas@....com,
        christian@...uner.io, cyphar@...har.com, dhowells@...hat.com,
        ebiederm@...ssion.com, fenghua.yu@...el.com, geert@...ux-m68k.org,
        gor@...ux.ibm.com, ink@...assic.park.msu.ru, jolsa@...hat.com,
        linux@...linux.org.uk, lkp@...el.com, mark.rutland@....com,
        mattst88@...il.com, minchan@...nel.org, mingo@...hat.com,
        monstr@...str.eu, namhyung@...nel.org, peterz@...radead.org,
        rth@...ddle.net, sargun@...gun.me, sfr@...b.auug.org.au,
        tony.luck@...el.com, akpm@...ux-foundation.org,
        alex.huangjianhui@...wei.com, zhongjubin@...wei.com,
        linux-kernel@...r.kernel.org, linux-s390@...r.kernel.org,
        clang-built-linux@...glegroups.com, kbuild-all@...ts.01.org,
        linux-mm@...ck.org
Subject: Re: [PATCH] s390: fix build error for sys_call_table_emu

On Thu, Jun 18, 2020 at 01:27:02PM +0200, Heiko Carstens wrote:
> On Thu, Jun 18, 2020 at 07:03:20PM +0800, Xiaoming Ni wrote:
> > Build error on s390:
> > 	arch/s390/kernel/entry.o: in function `sys_call_table_emu':
> > 	>> (.rodata+0x1288): undefined reference to `__s390_'
> > 
> > In commit ("All arch: remove system call sys_sysctl")
> >  148  common	fdatasync		sys_fdatasync			sys_fdatasync
> > -149  common	_sysctl			sys_sysctl			compat_sys_sysctl
> > +149  common	_sysctl			sys_ni_syscall
> >  150  common	mlock			sys_mlock			sys_mlock
> > 
> > After the patch is integrated, there is a format error in the generated
> > arch/s390/include/generated/asm/syscall_table.h:
> > 	SYSCALL(sys_fdatasync, sys_fdatasync)
> > 	SYSCALL(sys_ni_syscall,) /* cause build error */
> > 	SYSCALL(sys_mlock,sys_mlock)
> > 
> > There are holes in the system call number in
> >  arch/s390/kernel/syscalls/syscall.tbl. When generating syscall_table.h,
> > these hole numbers will be automatically filled with "NI_SYSCALL".
> > Therefore, delete the number 149 to fix the current compilation failure.
> >  Similarly, modify tools/perf/arch/s390/entry/syscalls/syscall.tbl.
> > 
> > Fixes: ("All arch: remove system call sys_sysctl")
> > Fixes: https://lore.kernel.org/linuxppc-dev/20200616030734.87257-1-nixiaoming@huawei.com/
> > Reported-by: kernel test robot <lkp@...el.com>
> > Signed-off-by: Xiaoming Ni <nixiaoming@...wei.com>
> > ---
> >  arch/s390/kernel/syscalls/syscall.tbl           | 1 -
> >  tools/perf/arch/s390/entry/syscalls/syscall.tbl | 1 -
> >  2 files changed, 2 deletions(-)
> > 
> > diff --git a/arch/s390/kernel/syscalls/syscall.tbl b/arch/s390/kernel/syscalls/syscall.tbl
> > index f17aaf6fe5de..bcaf93994e3c 100644
> > --- a/arch/s390/kernel/syscalls/syscall.tbl
> > +++ b/arch/s390/kernel/syscalls/syscall.tbl
> > @@ -138,7 +138,6 @@
> >  146  common	writev			sys_writev			compat_sys_writev
> >  147  common	getsid			sys_getsid			sys_getsid
> >  148  common	fdatasync		sys_fdatasync			sys_fdatasync
> > -149  common	_sysctl			sys_ni_syscall
> 
> This is not correct. It should be changed to:
> 
>    149  common	_sysctl			-				-
> 
> Otherwise the generated __NR__sysctl define will be lost from
> unistd.h, which should not happen. Looking at the link above it
> _looks_ like a similar mistake was done for arm64.

I think we're ok on arm64, since it's only the compat syscall table that
is being updated and we don't export compat uapi headers (rather, they
come from arch/arm/).

Will