Re: [PATCH] Revert "x86/tsc: Consolidate init code"

[Pasha Tatashin <Pavel.Tatashin@microsoft.com>]

Hi Ville,


The failure is surprising, because the commit is tiny, and almost does
not change the code logic.

From looking through the commit, the only functional difference this
commit makes is:

static_branch_enable(&__use_tsc) was called unconditionally from
tsc_init(), but after the commit only when tsc_khz == 0.

I wonder if on p3 static_branch_enable(&__use_tsc) fails to enable
early, when it supposed to? But, I would first try to make that
unconditional call again, and see if this fixes the problem, and then
figure out why it was not enabled when it was supposed to.

So, in tsc_init(void)

First try to add this one line back:

        cyc2ns_init_secondary_cpus();
-       static_branch_enable(&__use_tsc);


See if it fixes everything, and lets work from there.

Thank you,
Pavel

On 9/10/18 8:48 AM, Thomas Gleixner wrote:
> Ville,
> 
> On Mon, 10 Sep 2018, Ville Syrjala wrote:
> 
>> From: Ville Syrjälä <ville.syrjala@linux.intel.com>
>>
>> This reverts commit 608008a45798fe9e2aee04f99b5270ea57c1376f.
>>
>> It breaks wifi on my pentium 3 Fujitsu-Siemens Lifebook S6010
>> laptop. Scanning for APs doesn't seem to work most of the time,
>> and, even when it manages to find some APs it never manages to
>> authenticate successfully. dmesg is just littered with:
>> "wlan0: send auth to ... (try 1/3)
>>  wlan0: send auth to ... (try 2/3)
>>  wlan0: send auth to ... (try 3/3)
>>  wlan0: authentication with ... timed out"
> 
> I asked for that before and I really do not understand why you do not even
> make an attempt to report an issue first and allow the developers to work
> with you to figure out what exactly is the problem. All you do is to send
> an revert patch with a changelog which describes symptoms and probably
> breaks more than it cures. Not really helpful, really.
> 
> It's surely helpful to know that you bisected it to that commit and
> reverting it helps. Can you please provide more detailes information like
> dmesg of an good and a bad boot?
> 
> Thanks,
> 
> 	tglx
> 
> 
>