PFS3AIO v3.1 Final - Page 24

AMIGASYSTEM · 20 December 2021, 14:43

Quote:

Originally Posted by Torti-the-Smurf

Thank you Toni and everybody involved in making the greatest FileSystem for the Amiga imho.

But i have one little Question, that really want to know ... hmm, how i make it short...? ( i try )

I think it's the signal that the file has been copied, with PFS if you restart the system after copying files without waiting, on restart you will notice that the files have not been copied.

Toni Wilen · 20 December 2021, 17:44

Quote:

Originally Posted by Torti-the-Smurf

What is happening there ?
What is this 1 second after write HD LED flash and why even after "boot with no startup-sequence" ?

I guess it clears/resets journal state, even if there was nothing to do. Normally it would mark "transaction" as completed or as aborted (for example if file was only written partially because system crashed mid-write or was reset, all writes in this "transaction" are "lost")

AMIGASYSTEM · 20 December 2021, 19:31

Quote:

Originally Posted by AMIGASYSTEM

Hi Toni, many years ago on my A4000 Reale "if I remember well" with PFS3 I could delete the files contained in the .deldir, on WinUAE these files are protected and cannot be deleted, am I remembering wrong?

Can anyone verify on Real Amiga if this happens, thanks

Swe_Kryten2x4b · 20 December 2021, 19:38

Quote:

Originally Posted by AMIGASYSTEM

Can anyone verify on Real Amiga if this happens, thanks

It does. Just tried on my wb-disk with pfs 19.2 (A1200).

AMIGASYSTEM · 20 December 2021, 21:34

Thanks, so you confirm that the files of the ".deldir" can be deleted !

Swe_Kryten2x4b · 20 December 2021, 22:25

Quote:

Originally Posted by AMIGASYSTEM

Thanks, so you confirm that the files of the ".deldir" can be deleted !

Sorry, no. The other way around. They are protected by default.

Edited to add: the files in there have RWED as the flags, but both dopus4 and shell refuses to delete them nonetheless.

AMIGASYSTEM · 20 December 2021, 23:55

Ok then I remembered wrong, I had the doubt because with SFS (Smart File System) from the "recycled" files can be deleted safely, thanks for the info.

Torti-the-Smurf · 21 December 2021, 13:05

Quote:

Originally Posted by Toni Wilen

I guess it clears/resets journal state, even if there was nothing to do. Normally it would mark "transaction" as completed or as aborted (for example if file was only written partially because system crashed mid-write or was reset, all writes in this "transaction" are "lost")

Kool, thanks you Tony and AmigaSystem for those Infos

You guys are great

cyberhead97 · 19 April 2022, 15:16

Is pfsdoctor experimental available for download anywhere?

I need it to test a PFS3 partition created and formatted without using WinUAE.

Jope · 19 April 2022, 17:14

The link in the last paragraph of the first post still works.

cyberhead97 · 19 April 2022, 22:37

Quote:

Originally Posted by Jope

The link in the last paragraph of the first post still works.

Ah, thanks. I was focused on the link to www.winuae.net on one of the later pages which doesn't work anymore and didn't notice the first page had a working link.

I successfully ported c code from Toni's pfs3aio github repo to C# covering formatting a PFS3 partition and pfsdoctor experimental reports it's valid

.

Firestone · 26 August 2022, 17:18

What is the latest version of PFS3 btw?
On github it seems like "v19" according to https://github.com/tonioni/pfs3aio/b...ionhistory.doc

I can see that on my A3000 if I version DH0, i get "pfs3aio 19.2". I don't know where I got it, because the version on Aminet is v18.6 as far as I know?

Couldn't find any binaries of newer versions in this thread...?

EDIT: Oh no my bad... the 19.2 version is the one on Aminet, not 18.6

cyberhead97 · 30 October 2022, 22:49

I have recently been porting PFS3AIO source code to C# to later build tools for directly read and write files to and from img/hdf images and physical drives.

While testing PFS3 partitions with different block size (sectors in partition block) it seems as this is completely ignored in PFS3 inner workings.

The metadata area/reserved blocks uses a block size controlled by the size of the partition and the data area uses 512 byte block size regardless.

So it seems changing block size on a partition with PFS3 doesn't have any effect. Does anyone know if this is true?

thomas · 31 October 2022, 12:32

Quote:

Originally Posted by cyberhead97

Does anyone know if this is true?

It's known and documented. Just RTFM.

cyberhead97 · 31 October 2022, 15:15

Quote:

Originally Posted by thomas

It's known and documented. Just RTFM.

I haven't read the manual, so i kinda expected this answer

utri007 · 11 November 2022, 09:46

Seems that new compile of Alien Breed 3D II - TKG can't save game, when partition filesystem is PFS.

https://eab.abime.net/showthread.php...28#post1574428

Toni Wilen · 11 November 2022, 11:29

Sounds like pfs3 working as designed.

If pfs3 can't do journal flush at the end of "transaction" (which happens few seconds after last write), whole disk operation gets rolled back (sort of, it is not exactly working like a database)

Toni Wilen · 14 July 2023, 12:24

Some news: Last few days I have been implementing support for configurable logical block sizes (2048 and 4096). Currently pfs3 always uses 1024 byte block size for metadata and 512 for data blocks which may not be that optimal when used with solid state devices.

But because >1024 block sizes also require pfs3 metadata blocks to have matching size (2048 or 4096), "large partition mode" (>104G) also needs to be enabled which still isn't 100% validated: some kind of brute-force filesystem tester is still needed. (Technically you could still use 1024 byte metadata blocks but it would be waste of disk space)

This update does not require data block structure changes.

There is also another update (which was mostly made 1-2 years ago): if partition is "superfloppy" (single partition, no partition tables), formatted geometry is stored in pfs3 root block extension block and when partition is mounted, stored geometry automatically overrides current (possibly wrongly guessed) geometry.

alexh · 14 July 2023, 13:36

Most filesystems I test the SSDs we design with use something like a 4Kbyte minimum block size. We announce we support 512 byte block size but internally we are 16Kbytes and anything less than that is sub optimal (i.e. require read-modify-write). This is because a typical NAND flash page is 16Kbytes

mfilos · 14 July 2023, 15:08

That's fantastic news Toni! Thanks for your hard work!
So once the new version gets released would be able to change our block size from 1024 to 4096 without having to reformat our partitions?
I was under the impression that you couldn't change the block size on-the-fly (but I might be wrong).

26 August 2022, 17:18	#472
Firestone Registered User Join Date: Apr 2013 Location: Norway Posts: 258	What is the latest version of PFS3 btw? On github it seems like "v19" according to https://github.com/tonioni/pfs3aio/b...ionhistory.doc I can see that on my A3000 if I version DH0, i get "pfs3aio 19.2". I don't know where I got it, because the version on Aminet is v18.6 as far as I know? Couldn't find any binaries of newer versions in this thread...? EDIT: Oh no my bad... the 19.2 version is the one on Aminet, not 18.6 Last edited by Firestone; 26 August 2022 at 17:22. Reason: I was wrong...=)

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
PFS3aio vs SFS/02	White	support.WinUAE	10	13 April 2018 21:44
PFS3AIO Probs ?	Foul	support.FS-UAE	0	21 February 2018 22:49
PFS3AIO Questions	manic23	support.Hardware	26	22 April 2014 13:49
Pfs3aio ????	stingray76	support.Hardware	2	19 February 2014 09:15
FinalWriter 97; FinalCalc 1.04; Final Data 2.00 & Final Copy	aebrown-u	request.Apps	20	12 July 2009 11:19

20 December 2021, 21:34	#465
AMIGASYSTEM Registered User Join Date: Aug 2014 Location: Brindisi (Italy) Age: 70 Posts: 8,255	Thanks, so you confirm that the files of the ".deldir" can be deleted !

20 December 2021, 23:55	#467
AMIGASYSTEM Registered User Join Date: Aug 2014 Location: Brindisi (Italy) Age: 70 Posts: 8,255	Ok then I remembered wrong, I had the doubt because with SFS (Smart File System) from the "recycled" files can be deleted safely, thanks for the info.

19 April 2022, 15:16	#469
cyberhead97 Registered User Join Date: Feb 2016 Location: Denmark Posts: 333	Is pfsdoctor experimental available for download anywhere? I need it to test a PFS3 partition created and formatted without using WinUAE.

19 April 2022, 17:14	#470
Jope - Join Date: Jul 2003 Location: Helsinki / Finland Age: 43 Posts: 9,880	The link in the last paragraph of the first post still works.

30 October 2022, 22:49	#473
cyberhead97 Registered User Join Date: Feb 2016 Location: Denmark Posts: 333	I have recently been porting PFS3AIO source code to C# to later build tools for directly read and write files to and from img/hdf images and physical drives. While testing PFS3 partitions with different block size (sectors in partition block) it seems as this is completely ignored in PFS3 inner workings. The metadata area/reserved blocks uses a block size controlled by the size of the partition and the data area uses 512 byte block size regardless. So it seems changing block size on a partition with PFS3 doesn't have any effect. Does anyone know if this is true?

11 November 2022, 09:46	#476
utri007 mä vaan Join Date: Nov 2001 Location: Finland Posts: 1,664	Seems that new compile of Alien Breed 3D II - TKG can't save game, when partition filesystem is PFS. https://eab.abime.net/showthread.php...28#post1574428

11 November 2022, 11:29	#477
Toni Wilen WinUAE developer Join Date: Aug 2001 Location: Hämeenlinna/Finland Age: 49 Posts: 26,536	Sounds like pfs3 working as designed. If pfs3 can't do journal flush at the end of "transaction" (which happens few seconds after last write), whole disk operation gets rolled back (sort of, it is not exactly working like a database)

14 July 2023, 12:24	#478
Toni Wilen WinUAE developer Join Date: Aug 2001 Location: Hämeenlinna/Finland Age: 49 Posts: 26,536	Some news: Last few days I have been implementing support for configurable logical block sizes (2048 and 4096). Currently pfs3 always uses 1024 byte block size for metadata and 512 for data blocks which may not be that optimal when used with solid state devices. But because >1024 block sizes also require pfs3 metadata blocks to have matching size (2048 or 4096), "large partition mode" (>104G) also needs to be enabled which still isn't 100% validated: some kind of brute-force filesystem tester is still needed. (Technically you could still use 1024 byte metadata blocks but it would be waste of disk space) This update does not require data block structure changes. There is also another update (which was mostly made 1-2 years ago): if partition is "superfloppy" (single partition, no partition tables), formatted geometry is stored in pfs3 root block extension block and when partition is mounted, stored geometry automatically overrides current (possibly wrongly guessed) geometry.

14 July 2023, 13:36	#479
alexh Thalion Webshrine Join Date: Jan 2004 Location: Oxford Posts: 14,408	Most filesystems I test the SSDs we design with use something like a 4Kbyte minimum block size. We announce we support 512 byte block size but internally we are 16Kbytes and anything less than that is sub optimal (i.e. require read-modify-write). This is because a typical NAND flash page is 16Kbytes

14 July 2023, 15:08	#480
mfilos Paranoid Amigoid Join Date: Mar 2008 Location: Athens/Greece Age: 45 Posts: 1,978	That's fantastic news Toni! Thanks for your hard work! So once the new version gets released would be able to change our block size from 1024 to 4096 without having to reformat our partitions? I was under the impression that you couldn't change the block size on-the-fly (but I might be wrong).

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)