| 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
| ||
|
Date: Sun, 20 Dec 2020 17:03:57 +0100
From: Christian Brauner <christian.brauner@...ntu.com>
To: Rae Kim <rae.kim@...il.com>
Cc: Oleg Nesterov <oleg@...hat.com>, Leesoo Ahn <dev@...eel.net>,
linux-kernel@...r.kernel.org, Leesoo Ahn <lsahn@...eel.net>,
"Eric W. Biederman" <ebiederm@...ssion.com>,
Jens Axboe <axboe@...nel.dk>,
Peter Collingbourne <pcc@...gle.com>,
Zhiqiang Liu <liuzhiqiang26@...wei.com>
Subject: Re: [PATCH] signal: Don't init struct kernel_siginfo fields to zero
again
On Mon, Dec 21, 2020 at 12:43:05AM +0900, Rae Kim wrote:
>
> It looks like compiler optimization is smart enough to know that
> assigning zero is unnecessary after clear_siginfo() which is memset()
> under the hood. At least in my x86_64 machine, w/ or w/o this patch,
> there is no difference in final compiled machine code. (I've compared
> "objdump -d" results for "__send_signal()", "do_tkill()", and
> "collect_signal()")
>
> Wouldn't it be nicer to have more information for both human and
> compiler since it doesn't generate extra machine code?
I don't have a strong preference. But the name clear_siginfo() is pretty
obvious imho. Say a new field "foo" were added to siginfo. We would
almost certainly reject a patch that would add an extra info->foo = 0
into all places where siginfo is initialized as being unnecessary unless
there was severe potential for confusion which I don't think is the case
here when removing this in favor of just relying on clear_siginfo(). But
as I said I don't have a strong opinion. I've picked this up but I'm
happy to drop it if other maintainers agree with you.
Christian
>
> On Sun, Dec 20, 2020 at 03:21:35PM +0100, Oleg Nesterov wrote:
> > On 12/20, Leesoo Ahn wrote:
> > >
> > > clear_siginfo() is responsible for clearing struct kernel_siginfo object.
> > > It's obvious that manually initializing those fields is needless as
> > > a commit[1] explains why the function introduced and its guarantee that
> > > all bits in the struct are cleared after it.
> > >
> > > [1]: commit 8c5dbf2ae00b ("signal: Introduce clear_siginfo")
> > >
> > > Signed-off-by: Leesoo Ahn <lsahn@...eel.net>
> >
> > Acked-by: Oleg Nesterov <oleg@...hat.com>
> >
> >
> > > ---
> > > kernel/signal.c | 21 ---------------------
> > > 1 file changed, 21 deletions(-)
> > >
> > > diff --git a/kernel/signal.c b/kernel/signal.c
> > > index 5736c55aaa1a..8f49fa3ade33 100644
> > > --- a/kernel/signal.c
> > > +++ b/kernel/signal.c
> > > @@ -603,10 +603,7 @@ static void collect_signal(int sig, struct sigpending *list, kernel_siginfo_t *i
> > > */
> > > clear_siginfo(info);
> > > info->si_signo = sig;
> > > - info->si_errno = 0;
> > > info->si_code = SI_USER;
> > > - info->si_pid = 0;
> > > - info->si_uid = 0;
> > > }
> > > }
> > >
> > > @@ -1120,7 +1117,6 @@ static int __send_signal(int sig, struct kernel_siginfo *info, struct task_struc
> > > case (unsigned long) SEND_SIG_NOINFO:
> > > clear_siginfo(&q->info);
> > > q->info.si_signo = sig;
> > > - q->info.si_errno = 0;
> > > q->info.si_code = SI_USER;
> > > q->info.si_pid = task_tgid_nr_ns(current,
> > > task_active_pid_ns(t));
> > > @@ -1133,10 +1129,7 @@ static int __send_signal(int sig, struct kernel_siginfo *info, struct task_struc
> > > case (unsigned long) SEND_SIG_PRIV:
> > > clear_siginfo(&q->info);
> > > q->info.si_signo = sig;
> > > - q->info.si_errno = 0;
> > > q->info.si_code = SI_KERNEL;
> > > - q->info.si_pid = 0;
> > > - q->info.si_uid = 0;
> > > break;
> > > default:
> > > copy_siginfo(&q->info, info);
> > > @@ -1623,10 +1616,7 @@ void force_sig(int sig)
> > >
> > > clear_siginfo(&info);
> > > info.si_signo = sig;
> > > - info.si_errno = 0;
> > > info.si_code = SI_KERNEL;
> > > - info.si_pid = 0;
> > > - info.si_uid = 0;
> > > force_sig_info(&info);
> > > }
> > > EXPORT_SYMBOL(force_sig);
> > > @@ -1659,7 +1649,6 @@ int force_sig_fault_to_task(int sig, int code, void __user *addr
> > >
> > > clear_siginfo(&info);
> > > info.si_signo = sig;
> > > - info.si_errno = 0;
> > > info.si_code = code;
> > > info.si_addr = addr;
> > > #ifdef __ARCH_SI_TRAPNO
> > > @@ -1691,7 +1680,6 @@ int send_sig_fault(int sig, int code, void __user *addr
> > >
> > > clear_siginfo(&info);
> > > info.si_signo = sig;
> > > - info.si_errno = 0;
> > > info.si_code = code;
> > > info.si_addr = addr;
> > > #ifdef __ARCH_SI_TRAPNO
> > > @@ -1712,7 +1700,6 @@ int force_sig_mceerr(int code, void __user *addr, short lsb)
> > > WARN_ON((code != BUS_MCEERR_AO) && (code != BUS_MCEERR_AR));
> > > clear_siginfo(&info);
> > > info.si_signo = SIGBUS;
> > > - info.si_errno = 0;
> > > info.si_code = code;
> > > info.si_addr = addr;
> > > info.si_addr_lsb = lsb;
> > > @@ -1726,7 +1713,6 @@ int send_sig_mceerr(int code, void __user *addr, short lsb, struct task_struct *
> > > WARN_ON((code != BUS_MCEERR_AO) && (code != BUS_MCEERR_AR));
> > > clear_siginfo(&info);
> > > info.si_signo = SIGBUS;
> > > - info.si_errno = 0;
> > > info.si_code = code;
> > > info.si_addr = addr;
> > > info.si_addr_lsb = lsb;
> > > @@ -1740,7 +1726,6 @@ int force_sig_bnderr(void __user *addr, void __user *lower, void __user *upper)
> > >
> > > clear_siginfo(&info);
> > > info.si_signo = SIGSEGV;
> > > - info.si_errno = 0;
> > > info.si_code = SEGV_BNDERR;
> > > info.si_addr = addr;
> > > info.si_lower = lower;
> > > @@ -1755,7 +1740,6 @@ int force_sig_pkuerr(void __user *addr, u32 pkey)
> > >
> > > clear_siginfo(&info);
> > > info.si_signo = SIGSEGV;
> > > - info.si_errno = 0;
> > > info.si_code = SEGV_PKUERR;
> > > info.si_addr = addr;
> > > info.si_pkey = pkey;
> > > @@ -1934,7 +1918,6 @@ bool do_notify_parent(struct task_struct *tsk, int sig)
> > >
> > > clear_siginfo(&info);
> > > info.si_signo = sig;
> > > - info.si_errno = 0;
> > > /*
> > > * We are under tasklist_lock here so our parent is tied to
> > > * us and cannot change.
> > > @@ -2033,7 +2016,6 @@ static void do_notify_parent_cldstop(struct task_struct *tsk,
> > >
> > > clear_siginfo(&info);
> > > info.si_signo = SIGCHLD;
> > > - info.si_errno = 0;
> > > /*
> > > * see comment in do_notify_parent() about the following 4 lines
> > > */
> > > @@ -2506,7 +2488,6 @@ static int ptrace_signal(int signr, kernel_siginfo_t *info)
> > > if (signr != info->si_signo) {
> > > clear_siginfo(info);
> > > info->si_signo = signr;
> > > - info->si_errno = 0;
> > > info->si_code = SI_USER;
> > > rcu_read_lock();
> > > info->si_pid = task_pid_vnr(current->parent);
> > > @@ -3660,7 +3641,6 @@ static inline void prepare_kill_siginfo(int sig, struct kernel_siginfo *info)
> > > {
> > > clear_siginfo(info);
> > > info->si_signo = sig;
> > > - info->si_errno = 0;
> > > info->si_code = SI_USER;
> > > info->si_pid = task_tgid_vnr(current);
> > > info->si_uid = from_kuid_munged(current_user_ns(), current_uid());
> > > @@ -3833,7 +3813,6 @@ static int do_tkill(pid_t tgid, pid_t pid, int sig)
> > >
> > > clear_siginfo(&info);
> > > info.si_signo = sig;
> > > - info.si_errno = 0;
> > > info.si_code = SI_TKILL;
> > > info.si_pid = task_tgid_vnr(current);
> > > info.si_uid = from_kuid_munged(current_user_ns(), current_uid());
> > > --
> > > 2.26.2
> > >
> >
Powered by blists - more mailing lists