@gregkh @sima “eventually calling __get_free_pages”?
You have to realize that when the GFP_xyz flags were introduced - back in 1992 - “get_free_page()” was the only way to allocate memory.
So “GFP” wasn’t some odd internal thing. There was nothing else (ok, there was a very simply malloc() library on top of that “you can free and allocate one page” mode).
No “eventually” about it. It was the thing.
Some of you know today as π-day.
But the real insiders know that today is the 30th anniversary of the 1.0 release of Linux.
Instead of all this “spring forward” stuff, is it ok to instead just sleep in for an extra 23 hours?
Asking for a friend.
@monsieuricon I think we should do this by default on the security list.
Not for the levity, but just to counteract the “security is more important than anything else” vibes. Security researchers have this “the sky is falling” thing going on for everything, it’s very tiresome.
Being the responsible parents we are, we have Carbon Monoxide alarms in the house, because hey, it’s what you do. Right?
Of course, they have never gone off (knock wood), so you do tend to forget that they exist at all.
Well, yesterday one of those alarms decided that it needed to really remind us that it exists, and that it’s been ten years since we activated it. Because it’s now time to replace it.
Of course, nobody was home - except for the dog. Who is now traumatized by that beeping hell-box that suddenly decided that it was a good idea to tell everybody out of the blue that it needs replacing - at 95dB, just to make sure.
Kidde - I’m sure you could at least start out with just a mild chirp, instead of going full “the bark collar from hell” crazy. No?
@eemeli Don’t do it.
Yes, it’s funny to do, and maybe git has no problems with it. Maybe.
But all the visualization tools will be ugly if they work at all. Just look at
`git show --cc 2cde51fbd0f3`
in the kernel repo and that’s a paltry 66 parents. We have a number of other “big” octopus merges, but we’ve avoided anything even that big since.
So even if things work, you just get into issues like “my terminal window isn’t that wide”.
@monsieuricon I considered it, but came up empty. Nothing struck my fancy.
@flightofdragons I do lattes. So the whole “buy beans directly from a roaster, grind them at home, and a dual-boiler espresso maker” setup.
It’s a twice-daily five-minute ritual to make a good(ish) cup of coffee.
My wife doesn’t much care, so we have a Keurig for her. She doesn’t do milk drinks, and just wants regular black coffee without the hassle.
@axboe that ‘spin_trylock_irq()’ is still potentially quite expensive in the failure case.
The trylock itself is probably fine, since it does
int val = atomic_read(&lock->val);
if (unlikely(val))
return 0;
so it’s just a single read for the contended case (sill potentially causing cacheline bounces, of course, but not nearly as much as trying to get the lock). But the irq part means that we end up doing
local_irq_disable(); \
raw_spin_trylock(lock) ? \
1 : ({ local_irq_enable(); 0; }); \
so for the “I didn’t get the lock” case you end up with that extra “disable and re-enable interrupts”.
That’s all because we don’t make each spinlock variation make their own irq-safe version, so we wrap the trylock around the whole thing.
And that bfq code seems to make this worse by doing this whole thing twice, both for bfq_bic_try_lookup() and bfq_merge().
The locking around ioc_lookup_icq() in particular looks odd, in that the function itself claims to be RCU-safe, but then it also requires that q->queue_lock, making a shambles of the whole RCU thing.
Staycat.. Oh, look at that - power stayed on.
There’s real pleasure in waking up and being able to make oneself a good cup of coffee. Not the disgusting swill I’ve been making my wife on the stove when the power was out.
@tommythorn you don’t think it’s because with a fixed DRAM chip, the FPGA people in question don’t have to worry about all the nasty timing detection details (SPD etc)?
Power back.
Let’s all sing “99 bottles of beer on the wall” to see how far we can get this time before it turns off again.
@kentborg honestly, I can do both, and will in fact just switch from one to the other depending on who I’m talking to. I have no issues with °F / °C.
The point where metric wins big is when you have order-of-magnitude conversions. The whole inch/feet/mile thing still makes no sense to me. But temperature readings just don’t have that issue.
Staycation: day seven.
We had power back twice yesterday for a couple of hours each time. The weather and PGE were just teasing us, and I will not consider that a real break in the cold darkness and despair that is being without electricity.
But I still have battery power, and while it froze some more overnight and outside is now very slippery, it’s once more above freezing. So frozen pipes remain a thing of the past.
End result: I have a handful of pull requests left (most of which came in in the last few hours), and no reason to believe that I won’t close the merge window normally on Sunday.
@brauner it’s much more expensive to do unless it’s a dense development. Basically overhead cables are cheap and “good enough” for almost all use.
I’ve lived in this house for 20 years now, and this is the second time ever we’ve really had anything like this. So being without power for a few days isn’t worth the worry.
From what I’ve seen (admittedly limited), underground power in the US is pretty much the norm whenever you have any kind of half-way dense development - whether it’s a developer building several houses together, or a town or a city. But feeder lines to the area tend to be aerial.
And part of the issue is then obviously that the US tends to be a lot less dense than Europe. At least in (southern) Finland, 100km is generally considered to be a reasonably long trip.
In the US, it might be a daily commute.
Of course, my bias is showing. I was a city boy when I lived in Finland. I considered places in downtown Helsinki that I couldn’t walk to to be outside the so-called “susiraja” (literally “wolf border”, city-boy speak for “wilderness”).
That took about two and a half hours. I have corrected my profile name to match.
Power came back on again.
Will it stay on for the whole night and still be up tomorrow? Your guess is as good as mine.