Well, maybe the current gcc maintainers are being (more than?) difficult, but there could be others who are really interested in your patches. If you just put it up on sourceforge.net, then they should be able to find it without too much hassle.
Don't let all suffer because the current maintainers are distracted by other things, or worse.
Remember gcc before? Development got stuck until approx version 2.7.2, then some people got fed up, forked (egcs), and now the leaders of egcs took over from the older gcc guys. Maybe it's time for another fork, or maybe they'll wise up.
Maybe too, your patches aren't the right thing right now for the current gcc source tree and it's current development track, however if you keep it to yourself, it will be lost forever, and if you put it out there, others can pick it up, use it, improve it, and it then actually may have a change to be accepted later on (maybe it needs 'cleaning up', and somebody comes around sees the patches, cleans it up and gets it ready for submission. Even open source projects have project development such as coding standards, and code freezes you know, and the good thing with open source is, that if the current 'owners' fall asleep, people can take it away from them and there is nothing they can do about it).
Well, with dependency paths defining the minimum time for a loop, and because of the fact that there is more than one resource that the processor can use each clock, there are many cases that inserting an instruction in a loop does not result in any increased cycles, yet does reduce the register pressure in the loop.
I've used such tricks myself on other architectures with confirmed speed increases, and I'm pretty sure the same is true for the P4 & Athlon.
In the 'old' days, execution speed was just counting instructions and their cycle counts, then processors came out with one cycle per instruction, making it easier, and then processors with a variable amount of instructions per cycle, making it less straight-forward again.
Yes you can, just send it a signal 19 (SIGSTOP), and wake it up with a signal 18 (SIGCONT).
When you press ctrl-z on a terminal process, the same thing happens (try for example 'find/', then press ctrl-z, you will get the prompt back and 'find' will be sleeping, type 'fg' to wake up find).
Just return an error message. The application has to be able to deal with lost connections anyway.
Note that you can SIGSTOP a process, then it will be on hold, may even become completely swapped out. Then you can SIGCONT the same process to let it run again.
So you could send it a SIGSTOP and force it to swapout. That is just checkpointing until the next reboot... Of course you need more info to restore the process from the swap when the system reboots, but it's a start as to how to implement checkpointing.
Then why didn't you and some colleages lift the server and the UPS together (without switching it off), put it in a car, and drive to the nearest place with electricity before the UPS runs out.
Or if the server was too heavy/big to move, then run to the hardware store and buy a generator
Other answer:
If it's a linux machine, then build a mosix cluster, then you can migrate the process to a system that still has power (assuming not all systems in the cluster lost power...)
"It is in our nature to get what we want at the best price."
History teaches us a bit more complicated nature than that.
When you really look at the fundamentals, it's our nature to strive for power.
You begin a society empty, with nothing, no relationships, no laws, no regulations, nobody 'has the power': anarchy.
Then we discover that by forming cooperation groups, the individuals in the group get more power than those not in a group.
A number of those groups form and rule over a geographical region: countries are born.
The empowered elite of each country enjoys their power. The rest of the inhabitants are mere 'subordinates'. This is often called 'monarchy', or 'dictatorship'. Sometimes the empowered elite consists of three separate cooperating empowered groups: One has the economical power (land owners, kings, emperors, etc), another has the military power, and yet another has a phychological power to control the subordinates (church). In some countries these three powers are in balance, in others one of the powers is obviously dominant (think about it, still today there are examples for each of those situations).
That stays until the unpowered other inhabitants discouver the cooperation group. They revolt or otherwise take power away from the dominant elite.
Sometimes that goes wrong and the leaders of the revolt become the new elite (has happened often), but when that does not happen a perfect democracy is born.
Then, in that democracy, people discover that if they form cooperative groups......
You will be charged such that the seller maximizes his profits. All that other stuff doesn't matter.
If you have no alternative, and really need your connection, you will be charged through the nose. If there is much competition and you don't value the service very much, it will be cheap.
It's not nice, but the cable company is not trying to be nice, it's trying to get money, and more of it. We the consumers are just some nice addicted TCP/IP junkies, we get sick if we have to be without it. No, worse, we feel sick and angry when they threaten to take it away. The first shot is always the best stuff, and free or cheap. They will try to hook you on it when you're young.
...
Now that everybody is nice and down... Does anybody know a good joke?
< irony > I do close my eyes when a man in black holds up a pen in my face. </irony >
Hope these people realyze they shouldn't wear sneakers with the tinfoil hat, because sneakers isolate to ground, and when it's not grounded, the tinfoil hat won't do much more than a little bending of part of the field.
Oh, and btw, did they ever check if the tinfoil did anything to block magnetic fields? FYI it doesn't, for that you need (expensive) low-mu materials.
Do they know spark plugs in cars generate pulsing electric fields? Or are they conveniently insensitive to that type of electric field?
Did they surrect a high perimeter fence yet to block terrestrial local television stations?
Agreed, and if you want a different syntax, who's to stop you taking perl and writing a translation script. Ease of configuration is a 'yeah, whatever'-issue, the firewall admin shouldn't care, because a firewall doesn't need constant reconfiguration.
Really, for a firewall all that matters in the end is, which is more secure, reliable, and faster, for which there is no simple answer.
Linux computers are getting faster every day. So we should probably end up with less, more powerful servers. Instead we are seeing more and more servers. While there are many reasons for this trend (more services offered), the major issue is more related to security and administrative concerns.
Is it possible to split a Linux server into virtual ones with as much isolation as possible between each one, looking like real servers, yet sharing some common tasks (monitoring, backup, ups, hardware configuration,...) ?
Just like Microsoft used Madonna for their advertisements. That doesn't mean that Madonna is now an employed MS salesperson. MS marketing obviously liked the result of Madonna's creative work.
A company hiring/supporting Alan would get instant recognition for doing that, without having to fully agree with him, just acknowledging his output. If they're hiring him as executive officer, it's a different story, but otherwise it may me just a 'mascotte' or 'thank you'-thing.
I think most managers want to buy a license+cd+support from a single vendor. Easier to negotiate prices, easier to do upgrades, to trust that your support guy knows it all, etc. A lot of managers somehow have it in their head that only the creator of a software tool can really fix it if it is broken. Open source is not often completely understood by management.
"think about those eye-level brake lights to see how attrition works"
Yes I remember that. In the beginning, it was obligatory in the US, but _illegal_ in many european countries (illegal to install _any_ additional lights inside/on cars...)
Isn't that a clear message: one country saying "you have to install it, it's for your own safety", and the other "you can't install it, it's unsafe"...
Hope we get a more uniform regulation if ever this gets regulated too (personally, I prefer less regulation).
When you multiply 2 32-bit numbers and really need the full precision of the 64-bit result, yes, then you need some 64-bit registers. However, that does not mean you need to have a multiply instruction that accepts 64-bit inputs. Also, often you don't need more than 32 bits of the result. In that case a barrel shifter in the chip right after the multiplier would already give you what you want without needing the large and slow 64x64 multiplier in the chip.
On DSPs, you can often choose between 'integer mode' and 'fixed point mode'. In the former case they mean integer input values just like the CPU has, and in the latter case they mean values in the range [-1,1>, which places the decimal point 31 bits more towards the LSB. In 'fixed point mode', it's intuitively easier to stick with 32 bit precision if more precision is not needed.
Additionally, DSPs have 'MAC' instructions: "accum out = accumin + (in1*in2)". Often, the number of bits in the 'accum' registers is larger than the number of bits in the 'in1' and 'in2' multiply inputs. A 16-bit DSP often (always?) has at least 32 bit wide 'accum' registers, often more than that, with up to 4 or 8 overflow bits in some cases. You need the overflow bits when you use the MAC instruction repeatedly (which is done often in typical DSP algorithms). With 4 overflow bits, you can use the MAC instruction 14=16 times and be guaranteed you'll never overflow 'accum'.
Personally, I'd more prefer the CPUs to get more DSP features than a simple increase of 'bits'.
When an OS doesn't use a CPU feature (4M pages, using it just for the kernel doesn't count), that doesn't make the hacker better, it makes the OS not taking advantage of all CPU features (and therefore not running into the related CPU bugs...).
So this guy tried to do 4M pages, it didn't work well (he encountered the bug), and decided not to implement 4M pages at all. And for Linux, the guys just happened to implement 4M pages long before AMD created the processors with the bug.
Slashdot Mirror
Well, maybe the current gcc maintainers are being (more than?) difficult, but there could be others who are really interested in your patches. If you just put it up on sourceforge.net, then they should be able to find it without too much hassle.
Don't let all suffer because the current maintainers are distracted by other things, or worse.
Remember gcc before? Development got stuck until approx version 2.7.2, then some people got fed up, forked (egcs), and now the leaders of egcs took over from the older gcc guys. Maybe it's time for another fork, or maybe they'll wise up.
Maybe too, your patches aren't the right thing right now for the current gcc source tree and it's current development track, however if you keep it to yourself, it will be lost forever, and if you put it out there, others can pick it up, use it, improve it, and it then actually may have a change to be accepted later on (maybe it needs 'cleaning up', and somebody comes around sees the patches, cleans it up and gets it ready for submission. Even open source projects have project development such as coding standards, and code freezes you know, and the good thing with open source is, that if the current 'owners' fall asleep, people can take it away from them and there is nothing they can do about it).
Then open a project on sourceforge and put your patches there.
Well, with dependency paths defining the minimum time for a loop, and because of the fact that there is more than one resource that the processor can use each clock, there are many cases that inserting an instruction in a loop does not result in any increased cycles, yet does reduce the register pressure in the loop.
I've used such tricks myself on other architectures with confirmed speed increases, and I'm pretty sure the same is true for the P4 & Athlon.
In the 'old' days, execution speed was just counting instructions and their cycle counts, then processors came out with one cycle per instruction, making it easier, and then processors with a variable amount of instructions per cycle, making it less straight-forward again.
So I downloaded the Linux_cpu.zip
/lib/libm.so.6 (0x4002b000)
/lib/libc.so.6 (0x4004d000)
/lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000)
/lib/libm.so.6 (0x4002b000)
/lib/libc.so.6 (0x4004d000)
/lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000)
It contains a shared and a static library, and two binaries. Full of symbols, so I stip them
61124 bytes in libcxa.a
49356 bytes in libcxa.so.1
90380 bytes oblcpu_gcc
131736 bytes oblcpu_icc
$ ldd oblcpu_*
oblcpu_gcc:
libm.so.6 =>
libc.so.6 =>
oblcpu_icc:
libm.so.6 =>
libcxa.so.1 => not found
libc.so.6 =>
Ok, so the icc version needs the shared library to be loaded as well.
$ size libcxa.so.1 oblcpu_*
text data bss dec hex filename
22839 3008 124 25971 6573 libcxa.so.1
70563 15860 1923912 2010335 1eacdf oblcpu_gcc
93858 24236 1923768 2041862 1f2806 oblcpu_icc
Codesize for gcc: 70563 bytes
Codesize for icc: 93858 + 22839 = 116697
Hmm, that is a 65% increase in code size! Not to mention the increase in data size (can anybody say 'lookup table' or 'buffering'?)
Hmm... I wonder if they tried gcc optimizations such as '-funroll-all-loops'. Too bad they didn't provide the source so we could verify the results.
I got a free evaluation CD from Intel with the February issue of "Linux Magazine", so I'll be doing my own comparisons thank you.
Ah, fear of rejection. Where have I read that before? ;-)
Remember, your patch doesn't have to be 'accepted' to be used by you.
" If you could sleep processes "
/', then press ctrl-z, you will get the prompt back and 'find' will be sleeping, type 'fg' to wake up find).
Yes you can, just send it a signal 19 (SIGSTOP), and wake it up with a signal 18 (SIGCONT).
When you press ctrl-z on a terminal process, the same thing happens (try for example 'find
killall -19 myprogram
killall -18 myprogram
Just return an error message. The application has to be able to deal with lost connections anyway.
Note that you can SIGSTOP a process, then it will be on hold, may even become completely swapped out. Then you can SIGCONT the same process to let it run again.
So you could send it a SIGSTOP and force it to swapout. That is just checkpointing until the next reboot... Of course you need more info to restore the process from the swap when the system reboots, but it's a start as to how to implement checkpointing.
I'm sure there is more than one road to Rome.
Then why didn't you and some colleages lift the server and the UPS together (without switching it off), put it in a car, and drive to the nearest place with electricity before the UPS runs out.
Or if the server was too heavy/big to move, then run to the hardware store and buy a generator
Other answer:
If it's a linux machine, then build a mosix cluster, then you can migrate the process to a system that still has power (assuming not all systems in the cluster lost power...)
"...hence the term "borked" was invoked when..."
Hmm, interesting to learn that.
So it really has nothing to do with the muppetshow?
"It is in our nature to get what we want at the best price."
History teaches us a bit more complicated nature than that.
When you really look at the fundamentals, it's our nature to strive for power.
You begin a society empty, with nothing, no relationships, no laws, no regulations, nobody 'has the power': anarchy.
Then we discover that by forming cooperation groups, the individuals in the group get more power than those not in a group.
A number of those groups form and rule over a geographical region: countries are born.
The empowered elite of each country enjoys their power. The rest of the inhabitants are mere 'subordinates'. This is often called 'monarchy', or 'dictatorship'. Sometimes the empowered elite consists of three separate cooperating empowered groups: One has the economical power (land owners, kings, emperors, etc), another has the military power, and yet another has a phychological power to control the subordinates (church). In some countries these three powers are in balance, in others one of the powers is obviously dominant (think about it, still today there are examples for each of those situations).
That stays until the unpowered other inhabitants discouver the cooperation group. They revolt or otherwise take power away from the dominant elite.
Sometimes that goes wrong and the leaders of the revolt become the new elite (has happened often), but when that does not happen a perfect democracy is born.
Then, in that democracy, people discover that if they form cooperative groups......
lather, rinse.
"How do you suggest I be charged?"
You will be charged such that the seller maximizes his profits. All that other stuff doesn't matter.
If you have no alternative, and really need your connection, you will be charged through the nose. If there is much competition and you don't value the service very much, it will be cheap.
It's not nice, but the cable company is not trying to be nice, it's trying to get money, and more of it. We the consumers are just some nice addicted TCP/IP junkies, we get sick if we have to be without it. No, worse, we feel sick and angry when they threaten to take it away. The first shot is always the best stuff, and free or cheap. They will try to hook you on it when you're young.
...
Now that everybody is nice and down... Does anybody know a good joke?
Then what about peak hours? Or peak times, when there is big news and 90% of the customers want to go online to view it?
Sure, they'll claim the 'system overloaded', while in reality it's a designed-in overload...
< irony > I do close my eyes when a man in black holds up a pen in my face. < /irony >
Hope these people realyze they shouldn't wear sneakers with the tinfoil hat, because sneakers isolate to ground, and when it's not grounded, the tinfoil hat won't do much more than a little bending of part of the field.
Oh, and btw, did they ever check if the tinfoil did anything to block magnetic fields? FYI it doesn't, for that you need (expensive) low-mu materials.
Do they know spark plugs in cars generate pulsing electric fields? Or are they conveniently insensitive to that type of electric field?
Did they surrect a high perimeter fence yet to block terrestrial local television stations?
'400-470 MHz - Uncontrollable sexual urges, strange thoughts, dishonesty, attraction to interns, voices, balding, interest in congressional office'
You've got to be kidding!
Sounds like a crazed witch-hunt to me 'A Witch! Burn 'em! Burn 'em!'. Now where's camelot to bring the peasants to reason?
"I like to USE it not twek it all day, remember, some of us actually work."
Tweak all day? Some of us can create such a script in minutes... I suggest learning perl, it will save you a bundle of time.
Agreed, and if you want a different syntax, who's to stop you taking perl and writing a translation script. Ease of configuration is a 'yeah, whatever'-issue, the firewall admin shouldn't care, because a firewall doesn't need constant reconfiguration.
Really, for a firewall all that matters in the end is, which is more secure, reliable, and faster, for which there is no simple answer.
"Clustering under vserver scheme is pretty dumb, if it was in fact meant to be serious."
...) ?
...
The vserver scheme is the _inverse_ of a cluster, not meant at all for applications that you would otherwise run on a multi-node cluster.
Did you even go to the vserver homepage , and read the introduction of the introduction? ->
Linux computers are getting faster every day. So we should probably end up with less, more powerful servers. Instead we are seeing more and more servers. While there are many reasons for this trend (more services offered), the major issue is more related to security and administrative concerns.
Is it possible to split a Linux server into virtual ones with as much isolation as possible between each one, looking like real servers, yet sharing some common tasks (monitoring, backup, ups, hardware configuration,
We think so
Alan is not just an employee, he is an icon.
Just like Microsoft used Madonna for their advertisements. That doesn't mean that Madonna is now an employed MS salesperson. MS marketing obviously liked the result of Madonna's creative work.
A company hiring/supporting Alan would get instant recognition for doing that, without having to fully agree with him, just acknowledging his output. If they're hiring him as executive officer, it's a different story, but otherwise it may me just a 'mascotte' or 'thank you'-thing.
I think most managers want to buy a license+cd+support from a single vendor. Easier to negotiate prices, easier to do upgrades, to trust that your support guy knows it all, etc. A lot of managers somehow have it in their head that only the creator of a software tool can really fix it if it is broken. Open source is not often completely understood by management.
Maybe IBM should make an offer for RH too?
s/LSB/MSB/ obviously
i read it and "I've worked on' doesnt" translate into "It's in there, we're using it", it translates into "somebody took a look at it".
"think about those eye-level brake lights to see how attrition works"
Yes I remember that. In the beginning, it was obligatory in the US, but _illegal_ in many european countries (illegal to install _any_ additional lights inside/on cars...)
Isn't that a clear message: one country saying "you have to install it, it's for your own safety", and the other "you can't install it, it's unsafe"...
Hope we get a more uniform regulation if ever this gets regulated too (personally, I prefer less regulation).
"when you multiply, e.g. 100,000 * 100,000"
When you multiply 2 32-bit numbers and really need the full precision of the 64-bit result, yes, then you need some 64-bit registers. However, that does not mean you need to have a multiply instruction that accepts 64-bit inputs. Also, often you don't need more than 32 bits of the result. In that case a barrel shifter in the chip right after the multiplier would already give you what you want without needing the large and slow 64x64 multiplier in the chip.
On DSPs, you can often choose between 'integer mode' and 'fixed point mode'. In the former case they mean integer input values just like the CPU has, and in the latter case they mean values in the range [-1,1>, which places the decimal point 31 bits more towards the LSB. In 'fixed point mode', it's intuitively easier to stick with 32 bit precision if more precision is not needed.
Additionally, DSPs have 'MAC' instructions: "accum out = accumin + (in1*in2)". Often, the number of bits in the 'accum' registers is larger than the number of bits in the 'in1' and 'in2' multiply inputs. A 16-bit DSP often (always?) has at least 32 bit wide 'accum' registers, often more than that, with up to 4 or 8 overflow bits in some cases. You need the overflow bits when you use the MAC instruction repeatedly (which is done often in typical DSP algorithms). With 4 overflow bits, you can use the MAC instruction 14=16 times and be guaranteed you'll never overflow 'accum'.
Personally, I'd more prefer the CPUs to get more DSP features than a simple increase of 'bits'.
When an OS doesn't use a CPU feature (4M pages, using it just for the kernel doesn't count), that doesn't make the hacker better, it makes the OS not taking advantage of all CPU features (and therefore not running into the related CPU bugs...).
So this guy tried to do 4M pages, it didn't work well (he encountered the bug), and decided not to implement 4M pages at all. And for Linux, the guys just happened to implement 4M pages long before AMD created the processors with the bug.
Different history, all good hackers.
I've had 1G TB systems sometimes lockup/reboot XFree86 overnight, while it ran the screensaver. You think it may be related?