From: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>
To: speck@linutronix.de
Subject: [MODERATED] Re: [PATCH v5 09/11] TAAv5 9
Date: Mon, 7 Oct 2019 23:01:56 -0700 [thread overview]
Message-ID: <20191008060156.GG5154@guptapadev.amr> (raw)
In-Reply-To: <20191006170646.GA147859@kroah.com>
> > +static DEFINE_MUTEX(tsx_mutex);
>
> I think I asked this before, but in looking at the code I still can't
> figure it out. What exactly is this protecting?
>
> It looks like you want to keep only one "writer" out of the sysfs store
> function at a time, but:
>
> > +ssize_t hw_tx_mem_store(struct device *dev, struct device_attribute *attr,
> > + const char *buf, size_t count)
> > +{
> > + enum tsx_ctrl_states requested_state;
> > + ssize_t ret;
> > + bool val;
> > +
> > + ret = kstrtobool(buf, &val);
> > + if (ret)
> > + return ret;
> > +
> > + mutex_lock(&tsx_mutex);
> > +
> > + if (val) {
> > + tsx_user_cmd = TSX_USER_CMD_ON;
> > + requested_state = TSX_CTRL_ENABLE;
> > + } else {
> > + tsx_user_cmd = TSX_USER_CMD_OFF;
> > + requested_state = TSX_CTRL_DISABLE;
> > + }
> > +
> > + /* Current state is same as the reqested state, do nothing */
> > + if (tsx_ctrl_state == requested_state)
> > + goto exit;
> > +
> > + tsx_ctrl_state = requested_state;
> > +
> > + tsx_update_on_each_cpu(val);
> > +exit:
> > + mutex_unlock(&tsx_mutex);
>
> What I think you want to do is just protect the tsx_update_on_each_cpu()
> function, right?
Also I believe below two operations needs to be under a lock. Without
the lock if there are two writers and one is preempted in between these
operations there is a possibility that tsx_ctrl_state and TSX hardware
state could go out of sync.
tsx_ctrl_state = requested_state;
// 1st writer gets preempted here
// 2nd writer flips tsx_ctrl_state and writes to the MSR.
// 1st writer wakes up and only writes to the MSR
tsx_update_on_each_cpu(val);
// tsx_ctrl_state and hardware state would be different here.
Chances of this happening is rare but still a possibility. The lock
would prevent such a condition.
>
> So, you are locking _outside_ of a function call? That's a sure way to
> madness over time. If this function is so special that it can not be
> called multiple times at once, then put the lock _inside_ the function,
> right?
>
> Otherwise you could have other places call that function, and this
> single lock is not going to protect anything :(
Future caller from other places should also take the lock and also
update tsx_ctrl_state. Worth mentioning in the comment.
> And why is tsx_user_cmd needed, and global?
This was added because cmdline_find_option() uses __initdata and can't
be called inside tsx_init() without it being an __init function.
tsx_init() is in S3/S5 and cpu hotplug path so it can't be __init in its
current callsite. With tsx=auto not translating to any of the
tsx_ctrl_state state, I added tsx_user_cmd. Anyways, I will drop this.
I hope it is okay to move tsx_init() to identify_boot_cpu() to avoid
adding new enum and writing to globals from percpu function. This way
tsx_init() is only called by boot cpu and secondary cpus call
tsx_en/disable().
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index 6b25039aa8ae..a4ce9e3a622c 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -1576,6 +1575,8 @@ void __init identify_boot_cpu(void)
#endif
cpu_detect_tlb(&boot_cpu_data);
setup_cr_pinning();
+
+ tsx_init();
}
diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c
index b1d6c96f6b88..96039b5cda7c 100644
--- a/arch/x86/kernel/cpu/intel.c
+++ b/arch/x86/kernel/cpu/intel.c
@@ -762,7 +762,10 @@ static void init_intel(struct cpuinfo_x86 *c)
init_intel_misc_features(c);
- tsx_init(c);
+ if (tsx_ctrl_state == TSX_CTRL_ENABLE)
+ tsx_enable();
+ else if (tsx_ctrl_state == TSX_CTRL_DISABLE)
+ tsx_disable();
}
------------------
tsx_init() can now be __init and use cmdline_find_option().
void __init tsx_init(void)
{
char arg[20];
int ret;
if (!tsx_ctrl_is_supported())
return;
ret = cmdline_find_option(boot_command_line, "tsx", arg, sizeof(arg));
if (ret >= 0) {
if (!strcmp(arg, "on")) {
tsx_ctrl_state = TSX_CTRL_ENABLE;
} else if (!strcmp(arg, "off")) {
tsx_ctrl_state = TSX_CTRL_DISABLE;
} else if (!strcmp(arg, "auto")) {
if (boot_cpu_has_bug(X86_BUG_TAA))
tsx_ctrl_state = TSX_CTRL_DISABLE;
else
tsx_ctrl_state = TSX_CTRL_ENABLE;
} else {
tsx_ctrl_state = TSX_CTRL_DISABLE;
pr_info("tsx: invalid option, defaulting to off\n");
}
} else {
/* tsx= not provided, defaulting to off */
tsx_ctrl_state = TSX_CTRL_DISABLE;
}
if (tsx_ctrl_state == TSX_CTRL_DISABLE) {
tsx_disable();
setup_clear_cpu_cap(X86_FEATURE_RTM);
} else if (tsx_ctrl_state == TSX_CTRL_ENABLE) {
tsx_enable();
setup_force_cpu_cap(X86_FEATURE_RTM);
}
}
I am hoping it is okay to use setup_clear_cpu_cap() in tsx_init().
Thanks,
Pawan
next prev parent reply other threads:[~2019-10-08 6:07 UTC|newest]
Thread overview: 75+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-10-05 6:17 [MODERATED] [PATCH v5 00/11] TAAv5 0 Pawan Gupta
2019-10-05 6:26 ` [MODERATED] [PATCH v5 01/11] TAAv5 1 Pawan Gupta
2019-10-05 6:27 ` [MODERATED] [PATCH v5 02/11] TAAv5 2 Pawan Gupta
2019-10-05 6:28 ` [MODERATED] [PATCH v5 03/11] TAAv5 3 Pawan Gupta
2019-10-05 6:29 ` [MODERATED] [PATCH v5 04/11] TAAv5 4 Pawan Gupta
2019-10-05 6:30 ` [MODERATED] [PATCH v5 05/11] TAAv5 5 Pawan Gupta
2019-10-05 6:31 ` [MODERATED] [PATCH v5 06/11] TAAv5 6 Pawan Gupta
2019-10-05 6:32 ` [MODERATED] [PATCH v5 07/11] TAAv5 7 Pawan Gupta
2019-10-05 6:33 ` [MODERATED] [PATCH v5 08/11] TAAv5 8 Pawan Gupta
2019-10-05 6:34 ` [MODERATED] [PATCH v5 09/11] TAAv5 9 Pawan Gupta
2019-10-05 6:35 ` [MODERATED] [PATCH v5 10/11] TAAv5 10 Pawan Gupta
2019-10-05 6:36 ` [MODERATED] [PATCH v5 11/11] TAAv5 11 Pawan Gupta
2019-10-05 10:54 ` [MODERATED] Re: [PATCH v5 02/11] TAAv5 2 Borislav Petkov
2019-10-07 17:48 ` Pawan Gupta
[not found] ` <5d98396a.1c69fb81.6c7a8.23b1SMTPIN_ADDED_BROKEN@mx.google.com>
2019-10-05 21:43 ` [MODERATED] Re: [PATCH v5 03/11] TAAv5 3 Andy Lutomirski
2019-10-07 17:50 ` Pawan Gupta
[not found] ` <5d9839a4.1c69fb81.238e9.8312SMTPIN_ADDED_BROKEN@mx.google.com>
2019-10-05 21:45 ` [MODERATED] Re: [PATCH v5 04/11] TAAv5 4 Andy Lutomirski
[not found] ` <5d983ad2.1c69fb81.63edd.6575SMTPIN_ADDED_BROKEN@mx.google.com>
2019-10-05 21:49 ` [MODERATED] Re: [PATCH v5 09/11] TAAv5 9 Andy Lutomirski
2019-10-07 18:35 ` Pawan Gupta
[not found] ` <5d9838f1.1c69fb81.f1bab.d886SMTPIN_ADDED_BROKEN@mx.google.com>
2019-10-05 21:49 ` [MODERATED] Re: [PATCH v5 01/11] TAAv5 1 Andy Lutomirski
2019-10-06 17:40 ` Andrew Cooper
[not found] ` <5d983ad2.1c69fb81.e6640.8f51SMTPIN_ADDED_BROKEN@mx.google.com>
2019-10-06 17:06 ` [MODERATED] Re: [PATCH v5 09/11] TAAv5 9 Greg KH
2019-10-08 6:01 ` Pawan Gupta [this message]
2019-10-10 21:31 ` Pawan Gupta
2019-10-11 8:45 ` Greg KH
2019-10-21 8:00 ` Thomas Gleixner
2019-10-08 2:46 ` [MODERATED] Re: [PATCH v5 05/11] TAAv5 5 Josh Poimboeuf
2019-10-09 1:45 ` Pawan Gupta
2019-10-08 2:57 ` [MODERATED] Re: [PATCH v5 09/11] TAAv5 9 Josh Poimboeuf
2019-10-08 6:10 ` Pawan Gupta
2019-10-08 10:49 ` Jiri Kosina
2019-10-09 13:12 ` [MODERATED] Re: ***UNCHECKED*** [PATCH v5 08/11] TAAv5 8 Michal Hocko
2019-10-14 19:41 ` Thomas Gleixner
2019-10-14 19:51 ` [MODERATED] " Jiri Kosina
2019-10-14 21:04 ` [MODERATED] " Borislav Petkov
2019-10-14 21:31 ` Jiri Kosina
2019-10-15 8:01 ` Thomas Gleixner
2019-10-15 10:34 ` [MODERATED] Re: ***UNCHECKED*** " Michal Hocko
2019-10-15 13:06 ` Josh Poimboeuf
2019-10-15 13:10 ` Jiri Kosina
2019-10-15 15:26 ` Josh Poimboeuf
2019-10-15 15:32 ` Jiri Kosina
2019-10-15 19:34 ` Tyler Hicks
2019-10-15 20:00 ` Josh Poimboeuf
2019-10-15 20:15 ` Jiri Kosina
2019-10-15 20:35 ` Jiri Kosina
2019-10-15 20:54 ` Josh Poimboeuf
2019-10-15 20:56 ` [MODERATED] " Pawan Gupta
2019-10-15 21:14 ` Jiri Kosina
2019-10-15 23:12 ` Josh Poimboeuf
2019-10-15 23:13 ` [MODERATED] [AUTOREPLY] [MODERATED] [AUTOREPLY] Automatic reply: " James, Hengameh M
2019-10-16 4:52 ` [MODERATED] " Jiri Kosina
2019-10-16 5:05 ` Jiri Kosina
2019-10-21 21:15 ` Luck, Tony
2019-10-16 7:14 ` Josh Poimboeuf
2019-10-16 7:20 ` Jiri Kosina
2019-10-18 1:17 ` Ben Hutchings
2019-10-18 4:04 ` Pawan Gupta
2019-10-15 17:47 ` Borislav Petkov
2019-10-16 7:26 ` [MODERATED] Re: ***UNCHECKED*** " Jiri Kosina
2019-10-16 7:54 ` [MODERATED] Re: ***UNCHECKED*** " Michal Hocko
2019-10-16 9:23 ` [MODERATED] Re: ***UNCHECKED*** " Michal Hocko
2019-10-16 12:15 ` Thomas Gleixner
2019-10-16 18:34 ` [MODERATED] " Pawan Gupta
2019-10-18 0:14 ` Pawan Gupta
2019-10-21 8:09 ` Thomas Gleixner
2019-10-21 12:54 ` [MODERATED] Re: ***UNCHECKED*** " Michal Hocko
2019-10-21 20:01 ` [MODERATED] " Pawan Gupta
2019-10-21 20:33 ` Josh Poimboeuf
2019-10-21 20:34 ` Josh Poimboeuf
2019-10-21 20:33 ` Pawan Gupta
2019-10-21 23:01 ` Andrew Cooper
2019-10-21 23:37 ` Luck, Tony
2019-10-21 23:39 ` Andrew Cooper
2019-10-14 21:05 ` [MODERATED] Re: ***UNCHECKED*** " Michal Hocko
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20191008060156.GG5154@guptapadev.amr \
--to=pawan.kumar.gupta@linux.intel.com \
--cc=speck@linutronix.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).