From: "Jan Beulich" <JBeulich@suse.com>
To: Andrew Cooper <andrew.cooper3@citrix.com>
Cc: Wei Liu <wei.liu2@citrix.com>,
Xen-devel <xen-devel@lists.xen.org>,
Brian Woods <brian.woods@amd.com>,
Roger Pau Monne <roger.pau@citrix.com>
Subject: Re: [PATCH 7/9] x86/amd: Support context switching legacy SSBD interface
Date: Fri, 07 Dec 2018 03:25:07 -0700 [thread overview]
Message-ID: <5C0A4A830200007800203FA5@prv1-mh.provo.novell.com> (raw)
In-Reply-To: <67fb003a-1914-2051-2cd8-e1791e17a0cc@citrix.com>
>>> On 06.12.18 at 19:55, <andrew.cooper3@citrix.com> wrote:
> On 06/12/2018 10:51, Jan Beulich wrote:
>>
>>> + unsigned int socket = c->phys_proc_id, core = c->cpu_core_id;
>>> + struct ssbd_ls_cfg *cfg;
>>> + uint64_t val;
>>> +
>>> + ASSERT(cpu_has_legacy_ssbd);
>>> +
>>> + /*
>>> + * Update hardware lazily, as these MSRs are expensive. However, on
>>> + * the boot paths which pass NULL, force a write to set a consistent
>>> + * initial state.
>>> + */
>>> + if (*this_ssbd == disable && next)
>>> + return;
>>> +
>>> + if (cpu_has_virt_sc_ssbd) {
>>> + wrmsrl(MSR_VIRT_SPEC_CTRL,
>>> + disable ? SPEC_CTRL_SSBD : 0);
>>> + goto done;
>>> + }
>>> +
>>> + val = ls_cfg_base | (disable ? ls_cfg_ssbd_mask : 0);
>>> +
>>> + if (c->x86 < 0x17 || c->x86_num_siblings == 1) {
>>> + /* No threads to be concerned with. */
>>> + wrmsrl(MSR_AMD64_LS_CFG, val);
>>> + goto done;
>>> + }
>>> +
>>> + /* Check that we won't overflow the worse-case allocation. */
>>> + BUG_ON(socket >= ARRAY_SIZE(ssbd_ls_cfg));
>>> + BUG_ON(core >= ssbd_max_cores);
>> Wouldn't it be better to fail onlining of such CPUs?
>
> How? We've not currently got an ability to fail in the middle of
> start_secondary(), which is why the previous patch really does go an
> allocate the worst case.
smp_callin() very clearly has failure paths, and that's being
called out of start_secondary(). If you look there you'll notice
that it wasn't all that long ago that we've added a second
failure path here besides the HVM enabling one (which has been
there virtually forever).
>>> + cfg = &ssbd_ls_cfg[socket][core];
>>> +
>>> + if (disable) {
>>> + spin_lock(&cfg->lock);
>>> +
>>> + /* First sibling to disable updates hardware. */
>>> + if (!cfg->disable_count)
>>> + wrmsrl(MSR_AMD64_LS_CFG, val);
>>> + cfg->disable_count++;
>>> +
>>> + spin_unlock(&cfg->lock);
>>> + } else {
>>> + spin_lock(&cfg->lock);
>>> +
>>> + /* Last sibling to enable updates hardware. */
>>> + cfg->disable_count--;
>>> + if (!cfg->disable_count)
>>> + wrmsrl(MSR_AMD64_LS_CFG, val);
>>> +
>>> + spin_unlock(&cfg->lock);
>>> + }
>> Any reason for duplicating the spin_{,un}lock() calls?
>
> To avoid having a context-dependent jump in the critical region. Then
> again, I suppose that is completely dwarfed by the WRMSR.
If you're afraid of extra branches, how about
spin_lock(&cfg->lock);
cfg->disable_count -= !disable;
/* First sibling to disable and last sibling to enable updates hardware. */
if (!cfg->disable_count)
wrmsrl(MSR_AMD64_LS_CFG, val);
cfg->disable_count += disable;
spin_unlock(&cfg->lock);
(which I'd very much hope the compiler carries out with just
the single unavoidable branch in the middle)?
Jan
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
next prev parent reply other threads:[~2018-12-07 10:25 UTC|newest]
Thread overview: 51+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-12-03 16:18 [PATCH 0/9] xen/amd: Support for guest MSR_VIRT_SPEC_CTRL support Andrew Cooper
2018-12-03 16:18 ` [PATCH 1/9] x86/spec-ctrl: Drop the bti= command line option Andrew Cooper
2018-12-04 16:19 ` Jan Beulich
2018-12-03 16:18 ` [PATCH 2/9] x86/cpuid: Drop the synthetic X86_FEATURE_XEN_IBPB Andrew Cooper
2018-12-04 16:21 ` Jan Beulich
2018-12-03 16:18 ` [PATCH 3/9] x86/cpuid: Extend the cpuid= command line option to support all named features Andrew Cooper
2018-12-04 16:28 ` Jan Beulich
2018-12-06 12:52 ` Wei Liu
2018-12-03 16:18 ` [PATCH 4/9] x86/amd: Introduce CPUID/MSR definitions for per-vcpu SSBD support Andrew Cooper
2018-12-04 16:06 ` Woods, Brian
2018-12-05 16:39 ` Jan Beulich
2018-12-05 17:50 ` Andrew Cooper
2018-12-06 8:49 ` Jan Beulich
2018-12-06 18:35 ` Andrew Cooper
2018-12-03 16:18 ` [PATCH 5/9] x86/amd: Probe for legacy SSBD interfaces on boot Andrew Cooper
2018-12-04 16:15 ` Woods, Brian
2018-12-05 16:50 ` Jan Beulich
2018-12-05 17:09 ` Andrew Cooper
2018-12-06 8:53 ` Jan Beulich
2018-12-06 10:59 ` Jan Beulich
2018-12-28 16:30 ` Andrew Cooper
2019-01-04 8:58 ` Jan Beulich
2018-12-03 16:18 ` [PATCH 6/9] x86/amd: Allocate resources to cope with LS_CFG being per-core on Fam17h Andrew Cooper
2018-12-04 16:38 ` Woods, Brian
2018-12-05 16:57 ` Jan Beulich
2018-12-05 17:05 ` Andrew Cooper
2018-12-06 8:54 ` Jan Beulich
2018-12-06 18:46 ` Andrew Cooper
2018-12-06 19:25 ` Woods, Brian
2018-12-07 10:17 ` Jan Beulich
2018-12-03 16:18 ` [PATCH 7/9] x86/amd: Support context switching legacy SSBD interface Andrew Cooper
2018-12-04 20:27 ` Woods, Brian
2018-12-06 10:51 ` Jan Beulich
2018-12-06 18:55 ` Andrew Cooper
2018-12-07 10:25 ` Jan Beulich [this message]
2018-12-03 16:18 ` [PATCH 8/9] x86/amd: Virtualise MSR_VIRT_SPEC_CTRL for guests Andrew Cooper
2018-12-04 21:35 ` Woods, Brian
2018-12-05 8:41 ` Jan Beulich
2018-12-05 19:09 ` Andrew Cooper
2018-12-06 8:59 ` Jan Beulich
2018-12-06 19:41 ` Woods, Brian
2018-12-06 10:55 ` Jan Beulich
2018-12-03 16:18 ` [PATCH 9/9] x86/amd: Offer MSR_VIRT_SPEC_CTRL to guests Andrew Cooper
2018-12-06 10:57 ` Jan Beulich
2018-12-03 16:24 ` [PATCH 0/9] xen/amd: Support for guest MSR_VIRT_SPEC_CTRL support Jan Beulich
2018-12-03 16:31 ` Andrew Cooper
2018-12-04 9:45 ` Jan Beulich
2018-12-04 11:26 ` Andrew Cooper
2018-12-04 12:45 ` Jan Beulich
2018-12-04 13:41 ` Andrew Cooper
2018-12-04 14:07 ` Jan Beulich
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=5C0A4A830200007800203FA5@prv1-mh.provo.novell.com \
--to=jbeulich@suse.com \
--cc=andrew.cooper3@citrix.com \
--cc=brian.woods@amd.com \
--cc=roger.pau@citrix.com \
--cc=wei.liu2@citrix.com \
--cc=xen-devel@lists.xen.org \
/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).