Ok, then we'll look into it again. I wasn't trying to ignore you, but after your last post about the old HDD I concluded that its problem really had nothing to do with uLE, since you then said the following:
Originally Posted by katananja
Given that, I really had no reason to continue looking for some error that I never saw here myself.
Originally Posted by katananja
I'll try this some more myself, but exactly what do you mean by 'full' backup here ?
the HDD adapter is from other manufacturer and the HDD is a brand new 500GB IDE Western Digital Caviar Blue - WD5000AAKB, since is the 4rth time it corrupt the partition after a full MC backup with uLE beta(MC > HDD).
I assume that you are simply pushing square button to select all in the source root, and then browsing to the destination folder on mass: to issue a psuPaste command.
Have you tested if similar problems also affect the other two Paste commands ?
And most importantly, have you really tested conclusively that this problem is UNIQUE to the beta v4.39z ???
I don't see how its single change relative to stable v4.39 can affect the USB drivers, since they are unchanged. The only thing that could do it was if the new MC drivers somehow corrupt the IOP ram, thus trashing all drivers and their I/O buffers, but this should also crash the console which does not fit your description. Having the problem show its effects only after a turnoff points the finger at something else entirely IMO.
Which unfortunately rules me out. My most recent console is a 77K v15.
We need some one with a 90K console to test it before this beta goes stable.
Hmmm. If there is no error during the operation, and errors only appear after having turned off the PS2, this could mean that the physical write operations are not yet completed when you turn it off, some data still remaining in cache, either as handled by the drivers on the PS2 or handled internally in the drive itself. Either way such caching could easily result in trashed root directory sectors, effectively trashing the entire partition since the unharmed sectors are useless without a directory leading to them.
This time I did a 1Gb FAT32 partition for PS2 and the rest as NTFS for my PC backups, soon after the MC copy(select all > PSU Paste at the HDD) all looks fine at the end (the backup end without error), after you shutdown the PS2 for a few minutes and turn it on, when you go to MASS you see that screen full of junk instead of files and folder names.
This would correspond to the problem that any USB storage drive can suffer from if disconnected from a PC without first using the safe removal gadget, the difference being that it is then the PC OS which is doing the caching.
Is your USB drive powered from the PS2 or from an external power supply ?
That could affect the consequences of turn-off with unwritten cache data.
I have one myself which is supposedly Verbatim, but really a Samsung internally, and this is powered by the USB port so that PS2 turn-off leads to direct power-loss for the drive. I will test some full backups of 8MB MCs like you describe, to see if I can provoke errors similar to yours.
My guess is that this would be the occasions when you turn off the PS2 too soon after a transfer, so that some cached data is being written to directory sectors at the time of turn-off.
To make things worse the problem is aleatory, it doesn't happen every time in sequence, you can use one or ten times without any problem, but in a wrong day or a wrong time the partition get busted after the backup.
NB: I realize that the USB write-data rate of a PS2 is so low that any reasonable drive should be able to write the data without caching, or to empty that cache damned fast when an operation completes. But I also know that modern drives are not optimized for slow transfers, but rather the opposite, and using high-speed caching strategies may have unexpected effects when applied to a super-slow-speed transfer, such as leaving some sectors unwritten for a while in expectance of further writes to allow more efficient block access. And that is the factor that may allow you to turn off the PS2 before the cache is written... :(
Possibly my guess is wrong, but it is the only thing I can think of that would have this effect.
Using "host:" instead of FTP is even easier, so that is what I recommend users to do.
Have a look at the ProFTPD
, is open source and very light, the configuration is dead easy to do.
I am definitely not going to port another FTP server for use in uLE, since I am very doubtful about its need for any FTP server in the first place. The HOST protocol serves all my own PS2 network file transfer needs, and the only other solution I'd consider developing/porting would be a Samba interface rather than FTP. But only if it could be made bi-directional (which the Samba implementation of SMS is not).
Best regards: dlanor