From: Andi Kleen
Don't overflow array when the scripts directory is too large,
or the script file name is too long.
Signed-off-by: Andi Kleen
---
tools/perf/builtin-script.c | 8 ++--
tools/perf/builtin.h | 3 ++-
tools/perf/ui/browsers/scripts.c | 3 ++-
3 files changed
On Fri, Mar 01, 2019 at 06:58:32PM +0300, Alexey Budankov wrote:
Could do this as a follow up patch, but at some point the new
records need to be documented in Documentation/perf.data-file-format.txt
-Andi
> > + uname();
> > + if (!strcmp(uts.machine, session->header.env.arch) ||
> > + (!strcmp(uts.machine, "x86_64") &&
> > +!strcmp(session->header.env.arch, "i386")))
>
> why is this check and native_arch bool necessary?
> i386 data will be overed by arch/x86
This is so
> > +--samples=N::
> > + Save N individual samples for each histogram entry to show context in
> > perf
> > + report tui browser.
>
> maybe we could set some default value (50?)
50 wouldn't fit on the screen.
I turned it off by default intentionally because it will increase the
memory
From: Andi Kleen
When using the time sort key, add new context menus to run
scripts for only the currently selected time range. Compute
the correct range for the selection add pass it as the --time option to
perf script.
Signed-off-by: Andi Kleen
---
v2:
Use symbol_conf.time_quantum
v3:
Work
[Changes:
v3:
Fix compile problem on Fedora.
Rebase on latest tip. Now hopefully no missing patches.]
We currently have two ways to look at sample data in perf:
either use perf report to aggregate everything, or use
perf script to look at all individual samples.
Both ways are useful. Of course
From: Andi Kleen
Add a utility function to print nanosecond timestamps.
Signed-off-by: Andi Kleen
---
tools/perf/util/time-utils.c | 8
tools/perf/util/time-utils.h | 1 +
2 files changed, 9 insertions(+)
diff --git a/tools/perf/util/time-utils.c b/tools/perf/util/time-utils.c
index
From: Andi Kleen
Upcoming changes add timestamp output in perf report. Add a --ns
argument similar to perf script to support nanoseconds resolution
when needed.
Signed-off-by: Andi Kleen
---
v2:
Move flag into symbol_conf and change all users
---
tools/perf/Documentation/perf-report.txt | 3
From: Andi Kleen
The scripts menu traditionally only showed custom perf scripts.
Allow to run standard perf script with useful default options too.
- Normal perf script
- perf script with assembler (needs xed installed)
- perf script with source code output (needs debuginfo)
- perf script
From: Andi Kleen
Many workloads change over time. perf report currently aggregates
the whole time range reported in perf.data.
This patch adds an option for a time quantum to quantisize the
perf.data over time.
This just adds the option, will be used in follow on patches
for a time sort key
From: Andi Kleen
And one old option.
Signed-off-by: Andi Kleen
---
tools/perf/Documentation/tips.txt | 3 +++
1 file changed, 3 insertions(+)
diff --git a/tools/perf/Documentation/tips.txt
b/tools/perf/Documentation/tips.txt
index 849599f39c5e..4ec8107ed512 100644
--- a/tools/perf
From: Andi Kleen
The UI viewer for scripts output has a lot of limitations: limited size,
no search or save function, slow, and various other issues.
Just use 'less' to display directly on the terminal instead.
This won't work in gtk mode, but gtk doesn't support these
context menus anyways
From: Andi Kleen
Add a utility function to fetch executable code. Convert one
user over to it. There are more places doing that, but they
do significantly different actions, so they are not
easy to fit into a single library function.
Signed-off-by: Andi Kleen
---
tools/perf/util/Build
From: Andi Kleen
Now report can show whole time periods with perf script,
but the user still has to find individual samples of interest
manually.
It would be expensive and complicated to search for the
right samples in the whole perf file. Typically users
only need to look at a small number
From: Andi Kleen
perf script -F +insn was only working for PT traces because
the PT instruction decoder was filling in the insn/insn_len
sample attributes. Support it for non PT samples too on x86
using the existing x86 instruction decoder.
% perf record -a sleep 1
% perf script -F ip,sym,insn
From: Andi Kleen
Add a time sort key to perf report to display samples for
different time quantums separately. This allows easier
analysis of workloads that change over time, and also
will allow looking at the context of samples.
% perf record ...
% perf report --sort time,overhead,symbol
Commit-ID: 94816add0005595ea33fc8456519be582330401e
Gitweb: https://git.kernel.org/tip/94816add0005595ea33fc8456519be582330401e
Author: Andi Kleen
AuthorDate: Sun, 24 Feb 2019 07:37:19 -0800
Committer: Arnaldo Carvalho de Melo
CommitDate: Mon, 25 Feb 2019 10:58:28 -0300
perf tools
Commit-ID: 4b6ac811bce46c83811b83cdf87b41251596b9fc
Gitweb: https://git.kernel.org/tip/4b6ac811bce46c83811b83cdf87b41251596b9fc
Author: Andi Kleen
AuthorDate: Sun, 24 Feb 2019 07:37:12 -0800
Committer: Arnaldo Carvalho de Melo
CommitDate: Mon, 25 Feb 2019 10:58:07 -0300
perf script
On Wed, Feb 27, 2019 at 05:16:59PM +0100, Jiri Olsa wrote:
> On Wed, Feb 27, 2019 at 08:01:35AM -0800, Andi Kleen wrote:
> > > > Also available in
> > > > git://git.kernel.org/pub/scm/linux/kernel/git/ak/linux-misc.git
> > > > perf/streams-2
> > >
> > Also available in
> > git://git.kernel.org/pub/scm/linux/kernel/git/ak/linux-misc.git
> > perf/streams-2
>
> your post is missing this patch, it's only in the branch:
> perf tools: Add utility function to fetch executable
Because Arnaldo already merged it. But the branch is still based
on
Jiri Olsa writes:
>
> im still getting compile error the new branch:
>
> CC ui/browsers/hists.o
> ui/browsers/hists.c: In function ‘perf_evsel__hists_browse’:
> ui/browsers/hists.c:2567:8: error: ‘%s’ directive output may be truncated
> writing up to 63 bytes into a region of size
From: Andi Kleen
Upcoming changes add timestamp output in perf report. Add a --ns
argument similar to perf script to support nanoseconds resolution
when needed.
Signed-off-by: Andi Kleen
---
v2:
Move flag into symbol_conf and change all users
---
tools/perf/Documentation/perf-report.txt | 3
From: Andi Kleen
perf script -F +insn was only working for PT traces because
the PT instruction decoder was filling in the insn/insn_len
sample attributes. Support it for non PT samples too on x86
using the existing x86 instruction decoder.
% perf record -a sleep 1
% perf script -F ip,sym,insn
From: Andi Kleen
Also convert one existing user.
Signed-off-by: Andi Kleen
---
tools/perf/util/header.c | 12 +++-
tools/perf/util/util.c | 10 ++
tools/perf/util/util.h | 2 ++
3 files changed, 15 insertions(+), 9 deletions(-)
diff --git a/tools/perf/util/header.c b
From: Andi Kleen
Add a time sort key to perf report to display samples for
different time quantums separately. This allows easier
analysis of workloads that change over time, and also
will allow looking at the context of samples.
% perf record ...
% perf report --sort time,overhead,symbol
From: Andi Kleen
Many workloads change over time. perf report currently aggregates
the whole time range reported in perf.data.
This patch adds an option for a time quantum to quantisize the
perf.data over time.
This just adds the option, will be used in follow on patches
for a time sort key
From: Andi Kleen
When using the time sort key, add new context menus to run
scripts for only the currently selected time range. Compute
the correct range for the selection add pass it as the --time option to
perf script.
Signed-off-by: Andi Kleen
---
v2:
Use symbol_conf.time_quantum
[Changes:
Removed already merged patches.
Address review feedback, see individual patches.
Now compiles with gcc 8.
Some minor bug fixes and improvements.]
We currently have two ways to look at sample data in perf:
either use perf report to aggregate everything, or use
perf script to look at all
From: Andi Kleen
Add a utility function to print nanosecond timestamps.
Signed-off-by: Andi Kleen
---
tools/perf/util/time-utils.c | 8
tools/perf/util/time-utils.h | 1 +
2 files changed, 9 insertions(+)
diff --git a/tools/perf/util/time-utils.c b/tools/perf/util/time-utils.c
index
From: Andi Kleen
Now report can show whole time periods with perf script,
but the user still has to find individual samples of interest
manually.
It would be expensive and complicated to search for the
right samples in the whole perf file. Typically users
only need to look at a small number
From: Andi Kleen
And one old option.
Signed-off-by: Andi Kleen
---
tools/perf/Documentation/tips.txt | 3 +++
1 file changed, 3 insertions(+)
diff --git a/tools/perf/Documentation/tips.txt
b/tools/perf/Documentation/tips.txt
index 849599f39c5e..4ec8107ed512 100644
--- a/tools/perf
From: Andi Kleen
The UI viewer for scripts output has a lot of limitations: limited size,
no search or save function, slow, and various other issues.
Just use 'less' to display directly on the terminal instead.
This won't work in gtk mode, but gtk doesn't support these
context menus anyways
From: Andi Kleen
The scripts menu traditionally only showed custom perf scripts.
Allow to run standard perf script with useful default options too.
- Normal perf script
- perf script with assembler (needs xed installed)
- perf script with source code output (needs debuginfo)
- perf script
On Mon, Feb 25, 2019 at 11:40:45AM -0500, Sebastien Boisvert wrote:
>
>
> On 2019-02-24 10:37 a.m., Andi Kleen wrote:
> > From: Andi Kleen
> >
> > Upcoming changes add timestamp output in perf report. Add a --ns
> > argument similar to perf script to suppor
On Mon, Feb 25, 2019 at 01:56:15PM +0100, Jiri Olsa wrote:
> On Sun, Feb 24, 2019 at 07:37:22AM -0800, Andi Kleen wrote:
>
> SNIP
>
> > +static void hists__res_sample(struct hist_entry *he, struct perf_sample
> > *sample)
> > +{
> > + struct res_sample *r;
> for some reason can't see those items in menu
These one needs --samples 10 or similar.
It's off by default currently.
-Andi
> can't see the time ranges in the menu.. what's the path
> to get them listed?
You need to use --sort time,overhead,sym
Without time sorting there are no available ranges.
-Andi
From: Andi Kleen
Many workloads change over time. perf report currently aggregates
the whole time range reported in perf.data.
This patch adds an option for a time quantum to quantisize the
perf.data over time.
This just adds the option, will be used in follow on patches
for a time sort key
From: Andi Kleen
When using -F + syntax to add a field the existing defaults
are currently all marked user_set. This can cause errors when
some field is missing in the perf.data
This patch tracks the actually user set fields separately,
so that we don't error out in this case.
Before:
% perf
We currently have two ways to look at sample data in perf:
either use perf report to aggregate everything, or use
perf script to look at all individual samples.
Both ways are useful. Of course aggregation is useful
to quickly find the most expensive part of the code.
But sometimes a single
From: Andi Kleen
Also convert one existing user.
Signed-off-by: Andi Kleen
---
tools/perf/util/header.c | 12 +++-
tools/perf/util/util.c | 10 ++
tools/perf/util/util.h | 2 ++
3 files changed, 15 insertions(+), 9 deletions(-)
diff --git a/tools/perf/util/header.c b
From: Andi Kleen
The UI viewer for scripts output has a lot of limitations: limited size,
no search or save function, slow, and various other issues.
Just use 'less' to display directly on the terminal instead.
This won't work in gtk mode, but gtk doesn't support these
context menus anyways
From: Andi Kleen
Add a time sort key to perf report to display samples for
different time quantums separately. This allows easier
analysis of workloads that change over time, and also
will allow looking at the context of samples.
% perf record ...
% perf report --sort time,overhead,symbol
From: Andi Kleen
perf script -F +insn was only working for PT traces because
the PT instruction decoder was filling in the insn/insn_len
sample attributes. Support it for non PT samples too on x86
using the existing x86 instruction decoder.
% perf record -a sleep 1
% perf script -F ip,sym,insn
From: Andi Kleen
When using the time sort key, add new context menus to run
scripts for only the currently selected time range. Compute
the correct range for the selection add pass it as the --time option to
perf script.
Signed-off-by: Andi Kleen
---
tools/perf/ui/browsers/hists.c | 82
From: Andi Kleen
Add a utility function to print nanosecond timestamps.
Signed-off-by: Andi Kleen
---
tools/perf/util/time-utils.c | 8
tools/perf/util/time-utils.h | 1 +
2 files changed, 9 insertions(+)
diff --git a/tools/perf/util/time-utils.c b/tools/perf/util/time-utils.c
index
From: Andi Kleen
Now report can show whole time periods with perf script,
but the user still has to find individual samples of interest
manually.
It would be expensive and complicated to search for the
right samples in the whole perf file. Typically users
only need to look at a small number
From: Andi Kleen
The scripts menu traditionally only showed custom perf scripts.
Allow to run standard perf script with useful default options too.
- Normal perf script
- perf script with assembler (needs xed installed)
- perf script with source code output (needs debuginfo)
- perf script
From: Andi Kleen
Upcoming changes add timestamp output in perf report. Add a --ns
argument similar to perf script to support nanoseconds resolution
when needed.
Signed-off-by: Andi Kleen
---
tools/perf/Documentation/perf-report.txt | 3 +++
tools/perf/builtin-report.c | 1
On Thu, Feb 21, 2019 at 10:41:32AM +0100, Jiri Olsa wrote:
> And display the error message from removing
> the old data file:
>
> $ perf record ls
> Can't remove old data: Permission denied (perf.data.old)
> Perf session creation failed.
>
> Not sure how to make fail the rename (after we
On Thu, Feb 21, 2019 at 02:37:45PM +0100, Peter Zijlstra wrote:
> On Thu, Feb 21, 2019 at 08:12:25AM -0500, Prarit Bhargava wrote:
> > Users cannot disable multiple CPU features with the kernel parameter
> > clearcpuid=. For example, "clearcpuid=154 clearcpuid=227" only disables
> > CPUID bit
On Fri, Feb 15, 2019 at 08:56:02AM +, Wang, Wei W wrote:
> On Friday, February 15, 2019 12:32 AM, Andi Kleen wrote:
> >
> > > +static void intel_pmu_get_global_status(struct kvm_pmu *pmu,
> > > + struct msr_data *msr_info)
> >
> +static void intel_pmu_get_global_status(struct kvm_pmu *pmu,
> + struct msr_data *msr_info)
> +{
> + u64 guest_debugctl, freeze_lbr_bits = DEBUGCTLMSR_FREEZE_LBRS_ON_PMI |
> + DEBUGCTLMSR_LBR;
> +
> + if
> diff --git a/include/uapi/linux/perf_event.h b/include/uapi/linux/perf_event.h
> index 9de8780..ec97a70 100644
> --- a/include/uapi/linux/perf_event.h
> +++ b/include/uapi/linux/perf_event.h
> @@ -372,7 +372,8 @@ struct perf_event_attr {
> context_switch : 1, /*
> + case KVM_CAP_X86_GUEST_LBR:
> + r = -EINVAL;
> + if (cap->args[0] &&
> + x86_perf_get_lbr_stack(>arch.lbr_stack)) {
> + pr_err("Failed to enable the guest lbr feature\n");
Remove the pr_err. We don't want unprivileged users
On Mon, Feb 11, 2019 at 03:20:18PM +0100, Greg Kroah-Hartman wrote:
> 4.9-stable review patch. If anyone has any objections, please let me know.
>
> --
>
> From: Andi Kleen
>
> commit a7e3ed1e470116c9d12c2f778431a481a6be8ab6 upstream.
The patch
Commit-ID: 9b545c04abd4f7246a3bde040efde587abebb23c
Gitweb: https://git.kernel.org/tip/9b545c04abd4f7246a3bde040efde587abebb23c
Author: Andi Kleen
AuthorDate: Mon, 4 Feb 2019 14:23:30 -0800
Committer: Ingo Molnar
CommitDate: Mon, 11 Feb 2019 08:00:39 +0100
perf/x86/kvm: Avoid
> As my understanding, the microcode version for each stepping is independent
> and irrelevant. The 0x004e should be just coincidence.
> If so, I don't think X86_STEPPING_ANY is very useful.
>
> Andi, if I'm wrong please correct me.
Yes that's right. You cannot match microcode without
On Sun, Feb 03, 2019 at 08:05:53PM +0100, Jiri Kosina wrote:
> On Sun, 3 Feb 2019, Ben Hutchings wrote:
>
> > 3.16.63-rc1 review patch. If anyone has any objections, please let me know.
> >
> > --
> >
> > From: Jiri Kosina
> >
> > commit
Patches all look good to me.
Reviewed-by: Andi Kleen
-Andi
> Yeah, a loop stuck looks really scary inside an NMI handler.
> Should I just go ahead to send a patch to remove this warning?
> Or probably turn it into a pr_info()?
Not at this point. Would need to fix the PMU reset first to be
more selective.
-Andi
>
> Aside from all the missin {}, I'm fairly sure this is broken since this
> happens from NMI context. This can interrupt switch_mm() and things like
> use_temporary_mm().
So the concern is that the sample is from before the switch, and then
looks it up in the wrong page tables if the PMI
On Thu, Jan 31, 2019 at 01:28:34PM +0530, Ravi Bangoria wrote:
> Hi Andi,
>
> On 1/25/19 9:30 PM, Andi Kleen wrote:
> >> [Fri Jan 25 10:28:53 2019] perf: interrupt took too long (2501 > 2500),
> >> lowering kernel.perf_event_max_sample_rate to 79750
> &
Jiri Olsa writes:
>
> the patch adds check_eriod pmu callback.. I need to check if there's
> better way to do this, but so far it fixes the crash for me
>
> if you guys could check this patch, that'd be great
There's already a limit_period callback, perhaps that could
be extended. But ok, can do
On Tue, Jan 29, 2019 at 11:45:43AM +0100, Arnaldo Carvalho de Melo wrote:
> Em Mon, Jan 28, 2019 at 09:40:28AM +0300, Alexey Budankov escreveu:
> > The patch set implements runtime trace compression for record mode and
> > trace file decompression for report mode. Zstandard API [1] is used for
>
On Tue, Jan 29, 2019 at 12:05:36PM -0500, William Cohen wrote:
> Fix incorrect event names for the Load_Miss_Real_Latency metric for
> Cascadelake server in the same manner as commit 91b2b97025 for SKL/SKX.
Reviewed-by: Andi Kleen
>
> Signed-off-by: William Cohen
> ---
&g
> > also now it won't make sample for slave events
> > with zero value/period read
> >
> > note the patch needs to be split into more patches,
> > sending it all together for discussion over the solution
>
> any feedback on this one?
Looks good to me.
Reviewed-by: Andi Kleen
-Andi
> [Fri Jan 25 10:28:53 2019] perf: interrupt took too long (2501 > 2500),
> lowering kernel.perf_event_max_sample_rate to 79750
> [Fri Jan 25 10:29:08 2019] perf: interrupt took too long (3136 > 3126),
> lowering kernel.perf_event_max_sample_rate to 63750
> [Fri Jan 25 10:29:11 2019] perf:
Jerome Glisse writes:
>
> Right now this is more a temptative ie i do not know if i will succeed,
> in any case i can report on failure or success and discuss my finding to
> get people opinions on the matter.
I would just stop putting node/zone number into the flags. These
could be all handled
> + PERF_OUTPUT_CODE_PAGE_SIZE = 1UL << 32,
That won't work on 32bit. You need 1ULL
Also might want to audit that noone puts these flags into
an int.
-Andi
Commit-ID: 96167167b6e17b25c0e05ecc31119b73baeab094
Gitweb: https://git.kernel.org/tip/96167167b6e17b25c0e05ecc31119b73baeab094
Author: Andi Kleen
AuthorDate: Thu, 17 Jan 2019 11:48:34 -0800
Committer: Arnaldo Carvalho de Melo
CommitDate: Fri, 18 Jan 2019 09:53:07 -0300
perf script
> + /* Check the start address: needs to be page-aligned.. */
> +-if (start & ~PAGE_MASK)
> ++if (start & ~PAGE_MASK) {
> ++
> ++/*
> ++ * XXX Hack
> ++ *
> ++ * We re-use this error case to show case a cache load gadget:
> ++
> +static bool perf_evsel__should_skip(struct perf_evsel *evsel)
> +{
> + struct perf_event_attr *attr = >attr;
> + struct perf_evsel *leader = evsel->leader;
> +
> + return (leader != evsel) && !attr->freq && !attr->sample_freq;
> +}
> +
> static int process_sample_event(struct
gt;
> > But the value is still modified causing all sorts of inconsistencies:
> >
> > $ cat /proc/sys/kernel/perf_event_max_sample_rate
> > 10001
> >
> > This patch fixes the problem by moving the parsing of the value after
> > the test.
> >
> > Signed-off-by: Stephane Eranian
>
> Ping.
Reviewed-by: Andi Kleen
-Andi
From: Andi Kleen
perf script crashes currently when printing mixed trace points and other
events because the trace format does not handle events without trace
meta data. Add a simple check to avoid that.
% cat > test.c
main()
{
printf("Hello world\n");
}
^D
% gcc -g -o test
> +#ifdef CONFIG_X86_64
> +
> +#include
> +
> +.macro RDGSBASE opd
The caller can now use the assembler instructions directly, so the macros
are not needed anymore.
-Andi
> BTW: I am not sure that static-keys are much better. Their change also
> affects the control flow, and they do affect the control flow.
Static keys have the same problem, but they only change infrequently so usually
it's not too big a problem if you dump the kernel close to the tracing
On Tue, Jan 08, 2019 at 11:10:58AM +0100, Peter Zijlstra wrote:
> On Tue, Jan 08, 2019 at 12:01:11PM +0200, Adrian Hunter wrote:
> > The problem is that the jitted code gets freed from memory, which is why I
> > suggested the ability to pin it for a while.
>
> Then what do you tell the guy that
On Mon, Jan 07, 2019 at 10:48:38AM -0800, Jim Mattson wrote:
> On Mon, Jan 7, 2019 at 10:20 AM Andi Kleen wrote:
> >
> > > The issue is compatibility. Prior to your change, reading this MSR
> > > from a VM would raise #GP. After your change, it won't. That means
> The issue is compatibility. Prior to your change, reading this MSR
> from a VM would raise #GP. After your change, it won't. That means
> that if you have a VM migrating between hosts with kernel versions
> before and after this change, the results will be inconsistent. In the
No it will not
> Thanks for the explanations. I don’t think it would work (e.g., IRQs). I can
> avoid generalizing and just detect the "magic sequence” of the code, but let
> me give it some more thought.
If you ask me I would just use compiler profile feedback or autofdo (if your
compiler has a working
> Ok… I’ll try to think about another solution. Just note that this is just
> used as a hint to avoid unnecessary lookups. (IOW, nothing will break if the
> prefix is used.)
Are you sure actually?
The empty prefix could mean 8bit register accesses.
> > You're doing the equivalent of patching a
Nadav Amit writes:
I see another poor man's attempt to reinvent TSX.
> It is sometimes beneficial to have a restartable sequence - very few
> instructions which if they are preempted jump to a predefined point.
>
> To provide such functionality on x86-64, we use an empty REX-prefix
> (opcode
Nadav Amit writes:
>
> - Do we use periodic learning or not? Josh suggested to reconfigure the
> branches whenever a new target is found. However, I do not know at
> this time how to do learning efficiently, without making learning much
> more expensive.
FWIW frequent patching will likely
> Yes, but then what happens?
>
> Fast forward to, say, 2021. You're decommissioning all Broadwell
> servers in your data center. You have to migrate the running VMs off
> of those Broadwell systems onto newer hardware. But, with the current
> implementation, the migration cannot happen. So, what
Commit-ID: 61f611593f2c90547cb09c0bf6977414454a27e6
Gitweb: https://git.kernel.org/tip/61f611593f2c90547cb09c0bf6977414454a27e6
Author: Andi Kleen
AuthorDate: Mon, 19 Nov 2018 21:06:17 -0800
Committer: Arnaldo Carvalho de Melo
CommitDate: Fri, 28 Dec 2018 16:33:02 -0300
perf script
On Fri, Dec 28, 2018 at 11:47:06AM +0800, Wei Wang wrote:
> On 12/28/2018 04:51 AM, Andi Kleen wrote:
> > Thanks. This looks a lot better than the earlier versions.
> >
> > Some more comments.
> >
> > On Wed, Dec 26, 2018 at 05:25:38PM +0800, Wei Wang wrote:
Actually forgot one case.
In Arch Perfmon v4 the LBR freezing is also controlled through a GLOBAL_CTRL
bit.
I didn't see any code handling that bit?
-Andi
Thanks. This looks a lot better than the earlier versions.
Some more comments.
On Wed, Dec 26, 2018 at 05:25:38PM +0800, Wei Wang wrote:
> When the vCPU is scheduled in:
> - if the lbr feature was used in the last vCPU time slice, set the lbr
> stack to be interceptible, so that the host can
Masahiro Yamada writes:
> Revert the following 9 commits:
FWIW the -Wa additional also broke LTO builds because it doesn't really
support -Wa for individual files.
So I'm glad they got reverted.
-Andi
> In particular turning an address-dependency into a control-dependency,
> which is something allowed by the C language, since it doesn't recognise
> these concepts as such.
>
> The 'optimization' is allowed currently, but LTO will make it much more
> likely since it will have a much wider view
> You can always force disable SSB. In that case, all the child processes
> will have SSBD on.
Okay that sounds reasonable, given the below. Thanks.
-Andi
>
> >
> > Do you have a real use case where this behavior is a problem?
> >
> > -Andi
>
> Yes, we have an enterprise application partner
>"expecting a 'help' section of
> $min_conf_desc_length or more lines\n" . $herecurr);
> or maybe
>"please write a paragraph that describes
> the config symbol fully ($min_conf_desc_length or more lines)\n" . $herecurr);
>
On Wed, Dec 19, 2018 at 02:09:50PM -0500, Waiman Long wrote:
> With the default SPEC_STORE_BYPASS_SECCOMP/SPEC_STORE_BYPASS_PRCTL mode,
> the TIF_SSBD bit will be inherited when a new task is fork'ed or cloned.
>
> As only certain class of applications (like Java) requires disabling
> speculative
On Tue, Dec 18, 2018 at 05:16:20PM -0500, Steven Rostedt wrote:
> On Tue, 18 Dec 2018 14:13:38 -0800
> Andi Kleen wrote:
>
> > > Again, that's not the ftrace case. It doesn't care about more than one
> > > out of line instance. Thus, for this particular use, "u
On Tue, Dec 18, 2018 at 04:57:13PM -0500, Steven Rostedt wrote:
> Hmm, how does that work? When does LTO do its linker magic? Because the
> fentry/mcounts are added when the object is created. Are they removed
> if the compiler sees that it can be inlined? Or does LTO just compile
> everything in
On Tue, Dec 18, 2018 at 01:44:41PM -0800, Dave Hansen wrote:
> On 12/18/18 1:38 PM, Andi Kleen wrote:
> >> I misunderstood, you mean 32bit kernel, not 32bit machine. Theoretically
> >> 32bit
> >> kernel can use AVX512, but not sure if anyone use it like this.
>
> I misunderstood, you mean 32bit kernel, not 32bit machine. Theoretically 32bit
> kernel can use AVX512, but not sure if anyone use it like this.
> get_jiffies_64()
> includes jiffies_lock ops so not good in context switch. So I want to use raw
> jiffies_64 here. jiffies is a good candidate but
On Tue, Dec 18, 2018 at 10:19:32AM +0100, Peter Zijlstra wrote:
> On Mon, Dec 17, 2018 at 03:59:50PM -0800, Andi Kleen wrote:
> > BTW I have a user base for LTO and so far noone has reported any issues
> > like this.
>
> Because ordering issues are immediately obvious and eas
> I whittled it down to a small test case. It turns out the problem is
> caused by the "__optimize__("no-tracer")" atribute, which is used by our
> __noclone macro:
>
>
> # if __has_attribute(__optimize__)
> # define __noclone __attribute__((__noclone__,
>
501 - 600 of 19746 matches
Mail list logo