FatDog64 on early 2012 MacBook Pro, issues

For talk and support relating specifically to Puppy derivatives
Message
Author
User avatar
mavrothal
Posts: 3096
Joined: Mon 24 Aug 2009, 18:23

Re: continued

#16 Post by mavrothal »

Je55eah wrote: Regarding the same command, should it have taken effect as I ran it? Some time has passed now and the system has not cooled. I have not yet restarted because I am waiting for smartctl to finish but it seems to be stuck at 10% remaining. Moreover the change will not be retained across reboot until I get save files working again.
I believe you should reboot. Stop smartctl, reboot and run the echo command as soon as possible after that. Check again with 'top' that nothing else is using you CPU and run the other command to verify that no other hardware element is flooding the system with interrupts.

Regarding your HD, "find"(... :roll: ) DiskWarrior5 and run it first if you did not already. The more you poke with other tools the harder will be to recover anything from the HD, if is not a hardware failure.
Je55eah wrote: My thought was that the Mac seems to be unable to boot from CD, DVD, or USB in its current state that I should format the USB with GPT then do the second option
If this is true you may have totally different problems.
Not sure if you are a Mac user but booting from CD you need to press the 'C" key and to boot from anything else than the HD you need to press the 'option' key. If these fail you may want to reset the NVRAM with Command-Option-P-R. See more boot options here
== [url=http://www.catb.org/esr/faqs/smart-questions.html]Here is how to solve your[/url] [url=https://www.chiark.greenend.org.uk/~sgtatham/bugs.html]Linux problems fast[/url] ==

Je55eah
Posts: 28
Joined: Wed 16 Sep 2015, 12:25

#17 Post by Je55eah »

mavrothal,

See my reply above yours, you just got in there while I was typing it so I responded but it posted above you. I'll look for DiskWarrior5, but the good news is that I already saved all of her non-system files to other disks, and I made offline installers to augment the online installer which would still work if it were not for the hard drive errors. The disk can be treated as though it is already empty, in fact I ran badblocks -w on it for quite a few hours the other day. I also piped urandom into it for a good hour or so in the first case I was testing the drive and in the second I was hoping to clear the slate in case the partitioning tables were so messed up that fdisk or gdisk or disk utility were getting confused. Anyway thanks again, good night

Je55eah
Posts: 28
Joined: Wed 16 Sep 2015, 12:25

#18 Post by Je55eah »

This has been a week long crash course in Mac use. It hasn't been a great introduction to them. Yes, I tried both boot up keys combinations. Resetting the NVRAM was useful when the damn mac installers crashed, but it has not helped recently. The C key does cause the system to attempt to boot from the DVD, but it only works with EFI capable systems such as FatDog64. When the Hard drive was still holding some semblance of Macintosh OS X it would boot a windows install disc or a Puppy disc but it doesn't now. With the hard drive in it's current state the C key is not even required to boot the FatPup DVD. I could also boot Puppy from USB before if I option booted then inserted the usb, but not if I option booted with the USB already inserted. Now that the hard drive is essentially invisible I cannot boot puppy from USB. In all cases, those systems without EFI support trigger the macBook to put up a black screen and a message about inserting a bootable disc and pressing a key. As i said earlier, I can only surmise that the MacBook had some kind of chain boot feature for MBR that was on the hard drive.
Attachments
smart.txt.png
text file with fake extension. 10% remaining... probably meaningless
(9.9 KiB) Downloaded 161 times

User avatar
mavrothal
Posts: 3096
Joined: Mon 24 Aug 2009, 18:23

#19 Post by mavrothal »

Macs are totally different than windows (is FreeBSD in its core after all)
They have many builtin tools and safety valves, but I'm afraid you have removed all of them early on.
I'm using Mac since OS classic and all the problems I had with them (besides a couple of video card failures) were my doing :oops: But fortunately, never lost any data or drive.
== [url=http://www.catb.org/esr/faqs/smart-questions.html]Here is how to solve your[/url] [url=https://www.chiark.greenend.org.uk/~sgtatham/bugs.html]Linux problems fast[/url] ==

User avatar
mikeslr
Posts: 3890
Joined: Mon 16 Jun 2008, 21:20
Location: 500 seconds from Sol

Adding Ted Dog's "No Format Install" to the mix

#20 Post by mikeslr »

Hi Je55eah,

It's about 2 hours before my brain usually kicks in. But being awake and having nothing better to do, I took to reading threads I would ordinarily avoid as way beyond my knowledge. Read the foregoing as my justification for what may be a foolish response.

But, I was surprised that despite Ted Dog's posts to this thread, his "No-format install for UEFI based machines Win8 & MacTel" wasn't mentioned. http://murga-linux.com/puppy/viewtopic. ... 727#818727. One of the things it appears to be able to accomplish was the "loop-booting" of unicorn-6.2.1.91 which --as far as I recall-- is neither a 64-bit OS nor one specifically designed to run under UEFI systems. Note, however, what appears to be a "very slow boot" warning.

As the MacBook Pro 2012 was previously able to boot Pups other than 64-bit systems, I wondered if "No-format install" plus, say, Lucid Revisited might overcome the heating problem.

Just a probably half-baked thought.

mikesLr

User avatar
Ted Dog
Posts: 3965
Joined: Wed 14 Sep 2005, 02:35
Location: Heart of Texas

#21 Post by Ted Dog »

mavrothal is giving you a solid answer, and I will expand on it. Many distros boots will not work and the reason is the MacOS have a very weak support booting industry common BIOS or EFI. It has its own take on both, because of that, mixed mode hybrid DVDs and flashdrives rarely boot those contain BOTH BIOS and EFI methods and MacOS just gets confused.
FatDog64 has a different method ( and very non-standard if you think other distros are standard ) to hybrid boot. MacOS does not get confused by FatDog64 methodology so it is useful. The link given above for grub4EFI thread (mine) uses the EFI from FatDog64 to boot EFI ONLY, so macs are not confused.
The Benefit should be obvious read/write verses read only DVD. Back to BootCamp that idea uses the weak support and even weirder harddrive layout to make the impractical running Windows on Mac Hardware < to me its like taking a BMW and making run/look like a cheap Ford > There is a better MacOS 3rd party software for running windows at same time as MacOS forget the name.
You post like someone who is very 'techy' minded so reaching for complex answers to tough problems may be your Achilles heel. Do not over look the simple, you just have to copy files from the zip given above into a untouched generic fat formatted anything to get it to boot. Any other bootloader or MBR / GPT changes cause it to boot LESS reliably NOT improve it.
This simple method IS how EFI was setup to work. it works on both EFI based MacOS and Windows due to its design, nothing fancy just the simple.
FatDog64 'does' the magic handshake for us in its EFI folder for the complex crap Windows and supports the other lowlevel < roll of BootCamp > for the MacOS.
Check the other thread someone thankfully linked, I can never find my own threads. :oops: it has many odd ball configure setting that are easy to edit in the text based config. It has not been updated for a bit so either redownload the older versions or edit the config file with new iso names.

User avatar
Ted Dog
Posts: 3965
Joined: Wed 14 Sep 2005, 02:35
Location: Heart of Texas

#22 Post by Ted Dog »

is there an Intel kernel setting to not use drm?
like radeon.dpm=0 which allows my ATI based laptop to not use that ( drm seem very bad buggy and given up on with newest kernels ) that did drop my fan and heat issues I was having.

opps ignore.. eyesight still affected by stroke, drm != dpm

8)

Je55eah
Posts: 28
Joined: Wed 16 Sep 2015, 12:25

check in

#23 Post by Je55eah »

Thanks again everyone. Unfortunately I didn't find time today for this. I'll definitely be attempting the easiest solutions first. One bit of bad news is that the SDHC card which appeared to work briefly after reformatting it stopped working again on the MacBook before I went to bed yesterday.

jamesbond
Posts: 3433
Joined: Mon 26 Feb 2007, 05:02
Location: The Blue Marble

#24 Post by jamesbond »

Thanks guys for jumping in and helping around, taking care of the hard Mac-specific stuff :) I'll just answer the easy ones :)
I'm a bit embarrassed for missing that package. It looks promising.
Don't be, I built and uploaded it only after you reported the problem. It didn't exist before :) But I do hope it will help other Mac users too.
I did install mbpfan using Gslapt. I read that Gslapt is capable of dependency resolution, but I was not sure if it is thus configured on FatDog. Is it?
Yes, Fatdog's repo supports dependency resolution.
A third issue for this post, my razor tray is not displaying the temperature monitor or the battery monitor.
For that use pmcputemp (from 01micko). It's built-in, you need to go to Control Panel -> Desktop -> Desktop Startup Programs -> System startup programs and enable pmcputemp.sh.
Should I uninstall SeaMonkey? I know this would not save space on the disk, but the goal is to make a usable system for my friend.
I leave it to you. But if you do uninstall seamonkey, please install Firefox. Some of the programs depends on NSS and NSPR libraries, these are provided by either seamonkey or firefox.
I am running off of a DVD-R. The first two multisessions were saved successfully, but subsequent saves failed. How should I troubleshoot this. I have seen conflicting information in the documentation, often on the same page, about if DVD-R is untested, supported, or actually recommended.
I have never tested multisession with DVD-R; I only tested with DVD+RW and in fact that's the media that I recommends (since I don't see any benefit of DVD-R vs DVD+RW). You can test it by running the "savesession-dvd" from CLI (or there should be button "Saves to DVD" on your desktop when your're running in multisession mode); especially if you run the CLI version you can watch out for any errors.
Fatdog64 forum links: [url=http://murga-linux.com/puppy/viewtopic.php?t=117546]Latest version[/url] | [url=https://cutt.ly/ke8sn5H]Contributed packages[/url] | [url=https://cutt.ly/se8scrb]ISO builder[/url]

Je55eah
Posts: 28
Joined: Wed 16 Sep 2015, 12:25

update

#25 Post by Je55eah »

I have created a new DVD-R (no RW discs available) using v700. I'm currently working from the first multisession save file, so, so far so good. I attempted to use the SDHC card first, but the card reader was ultimately too unreliable. When it did work it was done with the EFI folder in the root of the flash drive and all of the other files from the DVD also copied to the root of the flash drive. I attempted to do it with only the EFI folder and the DVD iso but the loop mount menuentry didn't work. This may have been my fault. I entered that menuentry into refind.conf because that seemed to be the boot loader in use and it was the only conf file in the EFI folder. I did notice that refind in turn loaded grub so perhaps I was meant to modify grub.cfg and make a new iso or take that out of the iso as well...

As long as multisession continues to work, that doesn't matter.

Baring my heel a bit, I did proceed to test my theory by creating a FAT32 partition on a GPT partition table. The system booted that way, just the same, but there was no improvement in card reader performance.

I have the mbpfan module running, and the rc.local moification in place but the temperature has climbed in just the time it took for me to compose this. The fan does not seem to be blowing as hard as it could, and when I search for interrupts I see:

/sys/firmware/acpi/interrupts/gpe17: 580 disabled

I do wonder if that count of 580 is from before I rebooted as it does indicate gpe17 is disabled.

This is the temperature readout.

# sensors -f
coretemp-isa-0000
Adapter: ISA adapter
Physical id 0: +154.4°F (high = +188.6°F, crit = +221.0°F)
Core 0: +154.4°F (high = +188.6°F, crit = +221.0°F)
Core 1: +152.6°F (high = +188.6°F, crit = +221.0°F)

applesmc-isa-0300
Adapter: ISA adapter
Exhaust : 2119 RPM (min = 2000 RPM, max = 6200 RPM)
TA0P: +112.5°F
TB0T: +97.2°F
TB1T: +97.2°F
TB2T: +94.1°F
TC0E: +153.0°F
TC0F: +154.8°F
TC0J: +33.8°F
TC0P: +139.1°F
TC1C: +152.6°F
TC2C: +152.6°F
TCGC: +147.2°F
TCSA: +152.6°F
TCTD: +31.6°F
TCXC: +153.9°F
TG1D: +154.8°F
TM0P: +108.0°F
TM0S: +128.7°F
TPCD: +174.2°F
Th1H: +124.7°F
Ts0P: +89.1°F
Ts0S: +109.8°F

for comparison I found a cople links:
http://ubuntuforums.org/showthread.php?t=1308921

http://mac.appstorm.net/reviews/utiliti ... ure-gauge/

@mikeslr, I will be exploring your suggestion soon.

@jamesbond, Thank you, but I cannot find Control Panel -> Desktop -> Desktop Startup Programs -> System startup programs on v700. I did open the Add Plugins dialog from right clicking on the panel. I attempted to add the Sensors plugin, but nothing changed perceptibly.

You have all been great, I'm willing to keep working on this if you have more ideas, but I do not wish to burden any of you beyond your own level of interest.

Thanks again everyone,

ps. I attempted to use some bbcode in this post, but the whole post rendered invisible unless I either disabled bbcode or removed it, which I did.

update:
this is probably the system http://www.everymac.com/systems/apple/m ... specs.html
and the graphics card, intel HD graphics 4000, possibly runs this hot by design... http://hints.macworld.com/article.php?s ... 4054835249
and
http://www.intel.com/content/www/us/en/ ... sheet.html

User avatar
mikeslr
Posts: 3890
Joined: Mon 16 Jun 2008, 21:20
Location: 500 seconds from Sol

Further thoughts

#26 Post by mikeslr »

The objective of this thread is, despite a hard-drive failure, to keep a 2012 MacBook Pro as functional as possible for a few months until a replacement drive can be obtained.

Currently, the major problem is that while Fatdog64 works well, the computer is overheating. Ted Dog wondered whether overheating might have been the cause of the hard-drive failure. I'll state what might be obvious. If it was, merely replacing the hard-drive will ultimately result in the replacement drive's failure as well.

After an Htop report – see http://murga-linux.com/puppy/viewtopic. ... 4137#864137-- Je55eah asked whether it was possible to remove Seamonkey, Fatdog64's default browser, but Jamesbond advised that Seamonkey provides the NSS and NSPR libraries that are needed by other applications; if Seamonkey were removed it would have to be replaced by Firefox which also can provide those libraries.

Je55eah may not be completely aware that in Pups “removal

Je55eah
Posts: 28
Joined: Wed 16 Sep 2015, 12:25

update

#27 Post by Je55eah »

@mikeslr and others

Okay, v700 didn't work with the DVD-R multisession configuration either. After the first few successful saves every subsequent save failed. I cleared a 4GB USB flash drive and I'm back on v701 now.

After reading the post mikeslr referred me to I gained a better understanding of what Ted Dog was attempting to explain to me about booting from loop mounted iso files. I think it was described better at http://murga-linux.com/puppy/viewtopic. ... 727#818727 and I am using the EFI boot loading setup from Ted Dog's zip file.

Although it was not necessary I changed refind.conf to match the volume label of my usb drive. I also used gparted to set the volume label.

menuentry "Fatdog64 Linux" {
icon /EFI/boot/icons/os_fatdog.icns
volume KENNEL
loader /EFI/grub2/grub2.efi
}

I have modified the grub.conf file to read

menuentry "Start FatDog" {
loopback loop0 (hd0,msdos1)/Fatdog64-701.iso
linux (loop0)/vmlinuz waitdev=4 rootfstype=ramfs savefile=ram:multi:uuid:4248-738E:FatDogMac
initrd (loop0)/initrd
}

I had to take shot in the dark a bit. It is well documented that if a folder is specified that saves will not be done in a multisession style layered file systems. It would be nice if the folder could be specified as a folder to contain the dynamically named save files.
I tried savefile=ram:multi:uuid:4248-738E:/layers/FatDogMac but of course it didn't work. I have specified the uuid and saves are working but an accurate reading of http://distro.ibiblio.org/fatdog/web/fa ... tions.html doesn't indicates that this is an option with multi so perhaps /dev/sdb1/ is being selected by default. I would like to use the uuid to reduce the chance of saving to the wrong drive if another disk is attached.

I also specified rootfstype=ramfs which may be redundant and is not listed as an option in the boot-options-faq but I used it because it is used in grub.cg on the FatDog iso itself. Consequently, perhaps, I now have two icons in the upper right corner of the desktop: "Save RAM layer" and "Save session". What is the difference?

Following up on mikeslr's advise I also tried booting a few of the other puppy isos including racy and wary 5.5 but I was overwhelmed by options when attempting to find "lucid revisited." Is it here? http://murga-linux.com/puppy/viewtopic.php?t=90461

Nonetheless, only FatDog64 boots with the current configuration. Error messages indicate that various things are not found or loaded with the other pup isos. I used waitdev=20 to no effect so either they do not have the waitdev opton, which seems to be confirmed at http://puppylinux.org/wikka/BootParameters, or they will not work without some kind of built in support for efi. Help?

Is this grub.cfg file the configuration file for grub2? I see 2.00 on the top of the screen, but I had gotten the impression from previous experience that this file was not meant to be manually configured. Honestly I prefer it this way. I much prefer grub over grub2 . What is that boot menue that appears briefly between refind and grub 2.0? It has some blue in it. Also, I see another boot screen that asks for a specific grub.cfg or defaults to searching and finding the one we are configuring. Should I set this up somewhere so it doesn't need to search? Are we chain booting 4 boot loaders?

I modified rc.local to disable gpe17 and I installed mbpfan. I modified mbpfan.conf to:

[general]
min_fan_speed = 2000 # default is 2000
max_fan_speed = 6200 # default is 6200
low_temp = 50 # try ranges 55-63, default is 63
high_temp = 56 # try ranges 58-66, default is 66
max_temp = 82 # do not set it > 90, default is 86
polling_interval = 7 # default is 7

which is helping. The fans are running up to about 3500 rpm now and the highest sensor temp is hovering around 70. This is acceptable, but still rather warm. I will probably make further adjustments to mbpfan.conf yet. I think that there is some other problem with the graphics processor still causing it to run hot. I may be wrong, but it looks like the Intel drivers which are installed are a few versions behind the driver described on Intel's website. I don't know if upgrading them would help or how to do that. I also noticed that the libva and libvdpau modules are not installed. Is there any harm in trying them?

playing around with /sys/devices/platform/applesmc.768/fan1_manual produced some interesting results. mbpfan changes this to 1 (true) and seems to require it to be thus set. On the other hand when set to 0 the fans did something once.. /sys/devices/platform/applesmc.768/fan1_output can be manipulated. I haven't worked out all the details yet.

@jamesbond Since I am back on v701 I was able to successfully enable the razor temperature monitor. It isn't very useful however since, of all the sensors, it is showing the coolest ones. It also indicates that there are 4 processors, but this is apparently a dual core i7. I may be mistaken about this detail.

The battery monitor is still not visible. I read elsewhere that this may be an issue with a missing theme icon.

I think that is all for now.

jamesbond
Posts: 3433
Joined: Mon 26 Feb 2007, 05:02
Location: The Blue Marble

Re: update

#28 Post by jamesbond »

Je55eah wrote:I had to take shot in the dark a bit. It is well documented that if a folder is specified that saves will not be done in a multisession style layered file systems.
Correct. You either save in multisession mode or in savefolder/savefile, but not both.
It would be nice if the folder could be specified as a folder to contain the dynamically named save files. I tried savefile=ram:multi:uuid:4248-738E:/layers/FatDogMac but of course it didn't work.
This is supposed (and used to work). Apparently bitrot gets the better of it - I just tested - it no longer works, so I will look into it. PS: with "multi" mode, you don't need to use "ram" (use "direct" instead) unless you really need the double protection.
I have specified the uuid and saves are working but an accurate reading of http://distro.ibiblio.org/fatdog/web/fa ... tions.html doesn't indicates that this is an option with multi so perhaps /dev/sdb1/ is being selected by default. I would like to use the uuid to reduce the chance of saving to the wrong drive if another disk is attached.
UUID is supported. If you can get fatdog to boot, run "busybox blkid /dev/sdb1" (or whatever your device is" and note the UUID - that's the value that you need to use.
I also specified rootfstype=ramfs which may be redundant and is not listed as an option in the boot-options-faq but I used it because it is used in grub.cg on the FatDog iso itself.
This is to run Fatdog on memory constrained systems, allow Fatdog to claim as much RAM as needed to load itself. With standard Fatdog, you probably need this if your system has less than 1GB of RAM. The difference is between ability to boot Fatdog (or not); if you can boot without this option then it is not needed at all.
Consequently, perhaps, I now have two icons in the upper right corner of the desktop: "Save RAM layer" and "Save session". What is the difference?
That's because you specify "ram:multi" in you savefile parameter - you're running both the RAM layer and the Multi-layer. Replace it with "direct:multi" instead.
Nonetheless, only FatDog64 boots with the current configuration. Error messages indicate that various things are not found or loaded with the other pup isos. I used waitdev=20 to no effect so either they do not have the waitdev opton, which seems to be confirmed at http://puppylinux.org/wikka/BootParameters, or they will not work without some kind of built in support for efi. Help?
"waitdev" option is Fatdog specific. Other puppies have various level of supports implementing the same functionality (but with other names) - sorry I can't recall.
Is this grub.cfg file the configuration file for grub2?
Yes it is for grub 2.0
I see 2.00 on the top of the screen, but I had gotten the impression from previous experience that this file was not meant to be manually configured. Honestly I prefer it this way.
You *can* edit grub.cfg directly, if you know what you're doing. The "don't edit it manually" is becase in grub 2.0 they have split the configs into multiple "sub-parts", and then use a command to combine all these parts into a single grub.cfg; and you're supposed only to modify the "user-modifiable" parts - allowing the system to upgrade the other parts (which hasn't been edited by the users). Nevertheless, this doesn't make sense for Fatdog, so please feel free to edit it as you wish.
I much prefer grub over grub2 . What is that boot menue that appears briefly between refind and grub 2.0? It has some blue in it.
The first boot is "shim" - it checks for secure boot. Then the next one is "rEFInd". Then refind will boot grub2. Then grub2 will boot the kernel.
Also, I see another boot screen that asks for a specific grub.cfg or defaults to searching and finding the one we are configuring.
Can't remember, but you're probably referring to one of the items in the grub2 boot menu.
Should I set this up somewhere so it doesn't need to search? Are we chain booting 4 boot loaders?
You can for your specific needs. In general Fatdog can boot from multiple media so the only workable way is to search for it.
I modified rc.local to disable gpe17 and I installed mbpfan. I modified mbpfan.conf to:

[general]
min_fan_speed = 2000 # default is 2000
max_fan_speed = 6200 # default is 6200
low_temp = 50 # try ranges 55-63, default is 63
high_temp = 56 # try ranges 58-66, default is 66
max_temp = 82 # do not set it > 90, default is 86
polling_interval = 7 # default is 7

which is helping. The fans are running up to about 3500 rpm now and the highest sensor temp is hovering around 70. This is acceptable, but still rather warm.
Glad to hear that at least it helps a bit. Thanks for sharing so others with same problem can benefit from your fine-tuning of this config file.
I will probably make further adjustments to mbpfan.conf yet. I think that there is some other problem with the graphics processor still causing it to run hot. I may be wrong, but it looks like the Intel drivers which are installed are a few versions behind the driver described on Intel's website. I don't know if upgrading them would help or how to do that.
Your Mac is 2012; the drivers in 701 were from early 2015, so it should be good enough. That being said, there could be bugs in the version of the driver we have that got fixed in the newer version. You'll never know unless you read the changelogs :)
I also noticed that the libva and libvdpau modules are not installed. Is there any harm in trying them?
They're only used for playing video. You're having this problem without playing anything, so it probably won't help. But there is no harm installing them.
@jamesbond Since I am back on v701 I was able to successfully enable the razor temperature monitor. It isn't very useful however since, of all the sensors, it is showing the coolest ones. It also indicates that there are 4 processors, but this is apparently a dual core i7. I may be mistaken about this detail.
You can edit /root/.config/pmcputemp/pmcputemp.rc to point to the correct /sys/devices/* path for your temperature.
The battery monitor is still not visible. I read elsewhere that this may be an issue with a missing theme icon.
I have intermittent display issue with the battery monitor, but I got it right in about 9 of 10. Since I can't reproduce it reliably, this is an annoying problem to point point. But I'll see what I can do.

cheers!
Fatdog64 forum links: [url=http://murga-linux.com/puppy/viewtopic.php?t=117546]Latest version[/url] | [url=https://cutt.ly/ke8sn5H]Contributed packages[/url] | [url=https://cutt.ly/se8scrb]ISO builder[/url]

Je55eah
Posts: 28
Joined: Wed 16 Sep 2015, 12:25

progress

#29 Post by Je55eah »

Thanks again jamesbond,

I already found the ~/.config/pmcputemp/pmcputemprc and changed the contents to monitor what I can only guess is the sensor on the GPU. For the readers convenience pmcputemp is hosted and documented at https://github.com/01micko/pmcputemp You can't comment out the original contents of pmcputemprc but it doesn't matter because they are generated by the script itself which is in /usr/bin/. So delete the pre-existing line and replace it with no trailing line return. Mine now reads:

Code: Select all

/sys/devices/platform/applesmc.768/temp18_input
http://ark.intel.com/products/64893 indicates that this computer has four threads on two cores so whatever, good enough.

Another question I have, is about the save files. My clock is set wrong on this install, but I am afraid to correct it lest the new save predates the older save files. I could leave it off for a bunch of hours so that real time catches up, but I wondered if I can safely rename the save files to any arbitrary yet sequential earlier time?

On the same topic, can the existing save files be merged into one?

Also, I read that a person can avoid saving a session by removing the flash drive before shutting down. That's great, but what about after the fact? If I stick the drive in another computer and delete the last file will everything be okay, just regressed slightly? If I boot into the FatDog system itself and then delete the last save file, I imagine that would make a mess of things because at shutdown the save file would be expecting to build upon the earlier layer. Unless the layers are not based on prior knowledge. If the layer is built at shutdown from the current state compared to the last existing layer then maybe that is how I could consolidate the earlier layers... boot up, load the layers, delete the earlier layers, shutdown, save everything as a new layer.

Thanks,

oh and: What about the configurable 30 minute saves. I don't see this happening. That is fine, less disk wear, less clutter, but maybe risky.... but I just now noticed that a little yellow box at the top of the screen said it is saving. So I am just confused about how this works. Why would a RAM constrained system be allowed to use more RAM? Are changes saved to a single new save file on the disk starting from startup, continuing every 30 minutes, and finally ending at shutdown? What does the RAM layer save to? What about the session?

jamesbond
Posts: 3433
Joined: Mon 26 Feb 2007, 05:02
Location: The Blue Marble

Re: progress

#30 Post by jamesbond »

Je55eah wrote:Thanks again jamesbond,
No worries.
Another question I have, is about the save files. My clock is set wrong on this install, but I am afraid to correct it lest the new save predates the older save files. I could leave it off for a bunch of hours so that real time catches up, but I wondered if I can safely rename the save files to any arbitrary yet sequential earlier time?
Yes you can. The timestamp is used to determine the loading order (at bootup) and nothing else. It is save to change it to anything else.
On the same topic, can the existing save files be merged into one?
The savefile is a standard SFS file. A few industrious people on this forum have produced a few "SFS-merging" tool. I don't use them so I don't have the links handy but I'm sure you can find them.
Also, I read that a person can avoid saving a session by removing the flash drive before shutting down. That's great, but what about after the fact? If I stick the drive in another computer and delete the last file will everything be okay, just regressed slightly?
Yes. That's the point of multisession save, each "save" is a checkpoint, you can "regress" to another checkpoint by deleting (or temporarily ignoring) the last few checkpoints.
If I boot into the FatDog system itself and then delete the last save file, I imagine that would make a mess of things because at shutdown the save file would be expecting to build upon the earlier layer. Unless the layers are not based on prior knowledge. If the layer is built at shutdown from the current state compared to the last existing layer then maybe that is how I could consolidate the earlier layers... boot up, load the layers, delete the earlier layers, shutdown, save everything as a new layer.
You're very close; and your method to "merge" layer will work (indeed, that's what happens if your DVD/USB is full and you put on a new blank DVD/USB to start a new series). But you must make sure that you delete the *-base.sfs too. If in doubt, just copy these files into another USB first --- so if something goes wrong you can always try again.

At bootup, all the sessions are loaded into memory. This state is kept unchanged during operations (unless you click the "save session" icon to the desktop - but for simplicity let's don't touch that one yet). At shutdown, this previous state is saved into "*-base.sfs" IF AND ONLY IF there is no existing *-base.sfs. The changed files etc will be saved into a new *-save.sfs. On a normal operation, you will only have one *-base.sfs and multiple *-save.sfs on your USB/DVD.

Btw - I found that the bug that causes saving session to a path to fail. It is not bitrot, it's a genuine bug that has existed from 2012. Thanks for reporting it - it will get fixed on later updates.

cheers!
Fatdog64 forum links: [url=http://murga-linux.com/puppy/viewtopic.php?t=117546]Latest version[/url] | [url=https://cutt.ly/ke8sn5H]Contributed packages[/url] | [url=https://cutt.ly/se8scrb]ISO builder[/url]

jamesbond
Posts: 3433
Joined: Mon 26 Feb 2007, 05:02
Location: The Blue Marble

Re: progress

#31 Post by jamesbond »

Je55eah wrote:oh and: What about the configurable 30 minute saves. I don't see this happening. That is fine, less disk wear, less clutter, but maybe risky.... but I just now noticed that a little yellow box at the top of the screen said it is saving. So I am just confused about how this works. Why would a RAM constrained system be allowed to use more RAM? Are changes saved to a single new save file on the disk starting from startup, continuing every 30 minutes, and finally ending at shutdown? What does the RAM layer save to? What about the session?
There are two distinct modes of savefile operation on Fatdog. It's the savefile/savefolder and the multisession-save.
With savefile/savefolder, every change you make is immediately saved to disk.
With multisession-save, every change you make is buffered in a temporary location (in RAM), and only gets saved during either shutdown or when you click the "save session" icon.

On top of these two save modes, you can enable the "RAM" layer. This additional layer was originally conceived to save amount wear and tear if you use USB flash drive. The idea is, instead of directly saving changes to disk (=USB flash drive in this case), these changes get buffered into a RAM layer and only flushed to disk a) at shutdown, b) periodically (every 30mins by default) and c) if you click "save ram layer" icon.

In reality, this RAM layer operation has been used off label - mostly to enable "throwaway sessions" - simply disable the periodic save, and hack the system to disable "save at shutdown" (or at least ask for confirmation). Questions for "how not to save session at shutdown" shows up periodically in the Beginner's forum - about once a month, I think :lol: (e.g the latest one: http://murga-linux.com/puppy/viewtopic. ... 780#864780).

In other puppies, usage of RAM layer is automatic - it is enabled by default if it detects (or told) that the save media is flash memory. In Fatdog, RAM layer is never enabled by default. You have to tell it explicitly either to enable or disable the RAM layer. And you can use it in whatever save modes you use. Thus you can enable RAM layer for savefile/savefolder; and you can enable RAM layer for multisession save too. Note that 30 mins periodic save is a feature of the RAM layer.

That's why when you use "savefile=ram:multi" you've got two buttons; in this case you're actually operating with two RAM layers - one the RAM layer proper (with periodic save), and the temporary layer used for multisession mode; so the changed are actually buffered twice. First, in the RAM layer; and when you click "save RAM layer" (or when the periodic saving kicks in), it is moved to the temporary multisession layer. Only when you click "save session" the changes is saved in your media (usb/disk/dvd). Note however that only one copy of the data is kept at all times - either in the RAM layer or in the temporary multisession layer.

RAM layer is truly useful when you use it for savefile/savefolder; but questionable when you use multisession (in earlier post I actually suggest that you don't use it with multisession); it is there because RAM layer and save modes are independent of each other. Perhaps other can find a good reason to use it.

One warning: if you use "savefile=direct:multi", there is no RAM layer involved and there is no periodic save. Save happens when you shutdown or when you click the "save session" icon. If you want to do periodic save with multisession, you need to schedule the save job yourself (call the CLI "savesession-dvd"). Using RAM layer for this purpose will *not* work - the RAM layer period save is only from the RAM layer to whatever save mode used underneath it - and in the case of multisession, you still need to "savesession-dvd" to make it truly saved to your usb/dvd.

I hope this helps.
Fatdog64 forum links: [url=http://murga-linux.com/puppy/viewtopic.php?t=117546]Latest version[/url] | [url=https://cutt.ly/ke8sn5H]Contributed packages[/url] | [url=https://cutt.ly/se8scrb]ISO builder[/url]

Je55eah
Posts: 28
Joined: Wed 16 Sep 2015, 12:25

wow

#32 Post by Je55eah »

I really appreciate this. You probably get tired of explaining these things over and over, but I really understand this better now than I have before despite reading about it in other places.

I will be disabling the RAM layer, but as an exercise in understanding what you just presented I wonder if the system is booted as it is with both the RAM layer and the multisession RAM layer and the FatDog64 Event Manager RAM save interval is set to -0 (which, I gather, from the link you supplied, means never) then no changes would be saved at shutdown unless the save RAM layer button is pressed because the session layer would still be in the original state.

Another question about the save files. Can I copy & rename them to create a second profile with the same settings?

Changing the subject now, is sleep or hibernation possible with FatDog or Puppy? I know that MacBook Pros running Macintosh have a pretty slick standby feature. Since it takes a good two minutes to boot into FatDog a low power standby could be useful. I would probably hook into it and trigger a session save at standby as well.

Hmmn, now I am wondering about if the last save file is taken mid session and not at the moment just before poweroff completes will the operating system be in a strange state when it is started? This could happen if the battery died. How important are the things Linux does when it is shutting down?

User avatar
mikeslr
Posts: 3890
Joined: Mon 16 Jun 2008, 21:20
Location: 500 seconds from Sol

Mostly just following the discussion

#33 Post by mikeslr »

Hi again Je55eah and hopefully jamesbond,

Most of the discussion is far beyond my paygrade. But in answer to Je55eah's question regarding Lucid Revitalized, http://murga-linux.com/puppy/viewtopic. ... 518#739518 provides a list of the different Lucid Revitalized builds with explanations. However, links to the latest builds can be reached from this post, http://www.murga-linux.com/puppy/viewto ... 665#850665. These are essentially the same as on the list but with updated components.

However, as it appear that you've come a long way toward configuring FatDog64 into a serviceable OS, continuing along that path seems the best use of your time and effort.

Lest, however, the idea become lost during the pursuit of more pressing and fundamental issues, and as a heat issue still obtains to some extent, I'll mention again my query whether for browsing the internet 32-bit browsers via a 32bit emulation layer may require fewer computer resources --and generate less heat-- than 64-bit browsers? and If so, which 32bit emulation layer app in FatDog's repos would be appropriate?

mikesLr

jamesbond
Posts: 3433
Joined: Mon 26 Feb 2007, 05:02
Location: The Blue Marble

Re: wow

#34 Post by jamesbond »

Je55eah wrote:I really appreciate this. You probably get tired of explaining these things over and over, but I really understand this better now than I have before despite reading about it in other places.
No worries. Fatdog was designed for versatility, and some of the features indeed are rather obscure - even if they're documented. If you can get through with my explanation, I'm glad enough.
I will be disabling the RAM layer, but as an exercise in understanding what you just presented I wonder if the system is booted as it is with both the RAM layer and the multisession RAM layer and the FatDog64 Event Manager RAM save interval is set to -0 (which, I gather, from the link you supplied, means never) then no changes would be saved at shutdown unless the save RAM layer button is pressed because the session layer would still be in the original state.
Mostly correct - except for the fact that even if you press "save ram layer" button, it still won't be saved at all unless you also press "save session" button. The "RAM save interval" for most part is only used for the RAM layer, except that if its value is zero (="disable", you guessed right), save-session-at-shutdown is also disabled.
Another question about the save files. Can I copy & rename them to create a second profile with the same settings?
Yes.
Changing the subject now, is sleep or hibernation possible with FatDog or Puppy? I know that MacBook Pros running Macintosh have a pretty slick standby feature.
Yes, but it's a very finicky feature (in Linux in general, not only in Puppy/Fatdog). The main problem is driver support. Sleep (="suspend to RAM") is usually not problem, but resuming is. The driver needs to know that system is waking up from suspend and need to re-initialise itself - but some doesn't do this properly. These can be worked around but you need to know exactly which driver and hardware is problematic. Within this restriction, Puppy and Fatdog supports suspend; and while many can get it work, I regret that I myself can't get this to work on my laptop (my particular problem is a dark screen - everything comes back on resume except the LCD backlight).

I haven't built hibernation support to Fatdog. The reason is I can't even get suspend-to-RAM to work on my own laptop, so I'd expect hibernate will face a similar problem. Unless I'm mistaken there is no Puppy that supports hibernation too.
Since it takes a good two minutes to boot into FatDog a low power standby could be useful. I would probably hook into it and trigger a session save at standby as well.
Look into /etc/acpi/actions/suspend.sh to see show "suspend-to-RAM" is triggered. That script will be automatically executed if you close your laptop lid, but you can run it manually too.
Hmmn, now I am wondering about if the last save file is taken mid session and not at the moment just before poweroff completes will the operating system be in a strange state when it is started? This could happen if the battery died. How important are the things Linux does when it is shutting down?
I presume you mean what will happen if the system if forcefully shutdown before the save session is completed? (unplug power and/or battery dies)? The answer really depends on moments of failure. But generally speaking, the save will fail; the disk will be in inconsistent state. What happens on boot up also depends on the filesystem you use for saving. If you use FAT32, you could be in for a great shock - in the worse nightmare you could end up with cross-linked files and unfreed clusters, resulting in total data loss of that entire partition (that's what we get for using a 30-year old filesystem designed for single-user mode during an era where 8-bit CPUs reigned supreme). If you use modern Linux filesystem (ext3/ext4), the system will be able to recover and you will only miss your latest session. Thus, my suggestion is always: separate the "boot" partition and the "data/savefile" partition. The "boot" partition must be FAT32 if you're using EFI/UEFI (you know who to thanks to), and the "data" partition can be anything. The "boot" partition would never be written to so there is no risk of data loss there.

That being said, Ted Dog has a good point that many USB flashdrive firmware is preprogrammed to optimise for either FAT32 or NTFS; putting any other filesystem (or in even, the fact of re-formatting a flashdrive with) may confuse the firmware enough for it to fall back into "generic" mode which is a lot slower. You can still optimise this "generic" mode out but it probably won't beat the flashdrive firmware optimisation (unless you've got a really crappy flash drive).

So you can't eat cake and still have it, unfortunately. Alternatively, instead of USB flash drive, you can use USB harddisk, whose firmware doesn't play silly games like this. This is assuming your Mac can boot from such, though.
mikeslr wrote:Lest, however, the idea become lost during the pursuit of more pressing and fundamental issues, and as a heat issue still obtains to some extent, I'll mention again my query whether for browsing the internet 32-bit browsers via a 32bit emulation layer may require fewer computer resources --and generate less heat-- than 64-bit browsers? and If so, which 32bit emulation layer app in FatDog's repos would be appropriate?
Sorry mikeslr, I did miss the query. Thank you for pointing it to me again. The only proper way to address your question is by doing benchmark tests, which I haven't done. So what comes next is my best educated guess; but it is still a guess so take that with a (lot of) grain of salt.

I would say that 32-bit brower + 32-bit compat lib "vs" 64-bit browser --- all things being equal --- would have roughly the same power consumption, because equal work still needs to be done - page layouts, font rendering, etc. If you use Flash, I would reckon it will be slightly worse as 64-bit instructions are more efficient than 32-bit ones when it comes to decoding video etc though I have no idea how optimised is Flash for 64-bit, so really I'm plucking numbers out of thin air here.

For the 32-bit sfs to use - you can use 32bit-slacko6.sfs. This is actually a stripped down copy of Slacko6 (Slacko 5.97 I think - thanks 01micko!) with only the library files left. So if a 32-bit app can run on Slacko6, in theory it should be able to run on this too.
Fatdog64 forum links: [url=http://murga-linux.com/puppy/viewtopic.php?t=117546]Latest version[/url] | [url=https://cutt.ly/ke8sn5H]Contributed packages[/url] | [url=https://cutt.ly/se8scrb]ISO builder[/url]

Je55eah
Posts: 28
Joined: Wed 16 Sep 2015, 12:25

Re: wow

#35 Post by Je55eah »

jamesbond wrote:Mostly correct - except for the fact that even if you press "save ram layer" button, it still won't be saved at all unless you also press "save session" button. The "RAM save interval" for most part is only used for the RAM layer, except that if its value is zero (="disable", you guessed right), save-session-at-shutdown is also disabled.
But isn't the session always saved at shutdown even if the button is not pressed? Also, I gathered that zero will still save the RAM layer at shutdown, but negative zero will not.?
jamesbond wrote:I presume you mean what will happen if the system if forcefully shutdown before the save session is completed? (unplug power and/or battery dies)?
Actually, I mean what if the last save file is not interrupted, but is a result of pressing the save session button. It would be mid session. After that the battery could have died or the last save file could have been deleted.
jamesbond wrote:If you use FAT32, you could be in for a great shock - in the worse nightmare you could end up with cross-linked files and unfreed clusters, resulting in total data loss of that entire partition (that's what we get for using a 30-year old filesystem designed for single-user mode during an era where 8-bit CPUs reigned supreme). If you use modern Linux filesystem (ext3/ext4), the system will be able to recover and you will only miss your latest session.
I have often wondered if the benefits of ext3/ext4 journaling are preserved despite the underlying filesystem being FAT if the save file uses ext3/ext4. Likewise, in a virtual machine image file, wouldn't the VM filesystem retain it's fault tolerant features despite whatever filesystem contained the image? Is FAT worse than a raw disk?

Another question about save files. Since i am testing some boot parameters now I have created a grub2 profile that specifies no savefile=none and rootfstype=ramfs. is there a boot parameter that would disable the prompt to save and the second prompt to verify not saving at shutdown? I can see using something like that with or without a save file for testing and for guests.

I'm looking into modules that might affect the GPU now. This article is fairly informative http://www.phoronix.com/scan.php?page=a ... i915_power I can use dmesg, lsmod, and modinfo but I don't feel as if I am getting the whole picture. I am interested in using systool from sysfsutils but I don't know how to properly install this. http://linux-diag.sourceforge.net/Sysfsutils.html use case identified here:
http://serverfault.com/questions/62316/ ... ter-values

Another curiosity is https://lwn.net/Articles/444887/ which is already installed as intel_rapl but i am not sure how to use it yet.

This program is also interesting, but probably not a solution to my heat issue. http://lxr.free-electrons.com/source/dr ... ntel_ips.c

Speaking of backlights, the keyboard leds on the MacBook Pro are not illuminated. I have discoved that I can manually adjust them on a scale from 0-255 with a command such as:

Code: Select all

echo 255 > /sys/devices/platform/applesmc.768/leds/smc::kbd_backlight/brightness
Is there a ready made solution?

There are so many dark corners in Linux, it's amazing that anything works.

Did you say that FatDog is compatible with slackware? I thought it came from T2 which I gathered is a kind of semiautomated Linux From Scratch.

@mikeslr

What kernel boot parameters are you using with the non-FatDog isos?

Post Reply