Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.mobile.android > #143855

Re: Using Windows to make Android smoother

Cross-posted to 3 groups.

Show all headers | View raw

On Sat, 10/26/2024 6:10 AM, Arno Welzel wrote:
> Harry S Robins, 2024-10-24 20:00:
> 
> [...]
>> If a Windows full format doesn't fix bad sectors, what does it do to them?
> 
> It will verify every data block, so "bad" blocks can get recorded so
> they don't get used any longer.
> 
>> I always thought a format put a "jumper" so that bad sectors were ignored.
> 
> Yes - but this depends on the medium used. Flash storage media like SSDs
> have their own controller which does bad block mapping on their own and
> use spare blocks (usually a few percent of the capacity are reserved for
> this) as substitute for defect ones. However SD cards are quite "dumb"
> and 100% of the capacity is used for data - so bad blocks need to be
> recorded as part of the filesystem during "long" formatting.
> 

$BADCLUS can be updated at two times. By doing the "long format",
you can update $BADCLUS before files are put on a drive. By using the
scanning option in CHKDSK, you can mark off clusters via $BADCLUS
(presumably trashing a file at the same time, a file that was already
trashed so no big deal).

But with automatic sparing at the hard drive level, the need to
scan and add clusters to $BADCLUS is mostly removed. The only
time a disk gets a hard CRC error, is when that area of the
disk runs out of spares for repairs.

I had four CRC errors on a WD Blue in my Optiplex Refurb.
Re-writing the entire surface of the disk, flushed the errors.
There was an opportunity for the bad blocks to be spared out.
If I had done a $BADCLUS scan when the immediate problem
occurred, four clusters would be marked off as unusable,
and then there would be no need for the hard drive to
spare out the sector. But then, I have slightly less
space on the drive as a result.

You then have to ask yourself, what is the best practice for the
thing. Nothing springs to mind, except to say I would start
with an HDTune bad block scan, which would show the four bad sectors,
but without the side effect of using $BADCLUS. Then I would be in
a position to decide whether "washing and rinsing" the surface
was enough, or whether "digging divots" in the surface was needed.

When I checked, two of the CRC errors were in white space,
and two of the CRC errors were real files. By knowing what
the real files were, I could replace them. The white space
errors, it would not matter what happened there, so I had
only half the work to do (just dig up two files).

To work out what the files are, I use a copy of nfi.exe which
has LBA numbers. and then I can map an address to a file. Then
I can take a hex editor, open the "file" and notice the first
sector has stale data in it, the second sector has a CRC error.
This is the pattern you see on a "high fly" error. The head was
so far above the surface on the first sector, the mag field from
the write couldn't even couple into the surface.

You can push the entire partition into $BADCLUS. It has
the capacity to do that. if there is a pathological problem
via the controller, you could cause every file on a partition
to disappear.

In the past, a pathological situation for CHKDSK, occurs
when you pull the IDE ribbon cable half way off a drive. I
actually managed to do that one day, but I could feel that
I had bumped into something, and spotted it in time.
If you notice the disk has problems on the next boot,
you may be tempted to do a CHKDSK. It tries to rewrite stuff.
It modifies a hundred thousand things. Your disk is... trashed.
(None of the writes done, are good, because the cable is half off.)
It is for this sort of failing (an I/O issue), that CHKDSK
is the wrong tool. Other situations, the Microsoft declaration
that "stopping a CHKDSK run, causes no more damage than was there in
the first place", that statement can be true as long as the
I/O is in perfect working condition.

Summary: Firstly, a backup is your friend.
         HDTune can do a bad block scan, indicating your trouble situation.
         (I use more tools than this, this is just a canary run to prove trouble awaits.)
         ddrescue (gddrescue package in Ubuntu) can back up a damaged disk.
         [Normally you would use Macrium if the disk was undamaged, no CRCs]
         Since you have your backup now, you have my permission to run CHKDSK :-)
         Let the chainsaw at it.

         If $BADCLUS has a use in modern times, I don't really know where it
         fits. It was invented for a time when disks had less internal automation.
         Maybe it covers floppies well (our floppies at work used a similar philosophy).
         But generally speaking, it is my opinion that when the signs are
         that a sick disk drive is involved, you should make it personal and
         work at a lower level, to reduce the damage level. $BADCLUS, scanning
         disks for bad clusters, is just a bad economy. If your disk really
         sucks this badly, replace it. There are yoyos who will move a partition
         away from a CRC damaged area, but then those same individuals don't believe
         in backups, and they will be wearing that sad face when it dies entirely.

         The WD Blue 250GB in this story, got replaced. Cost $65 at the time for a WD Black 1TB.

   Paul

Back to comp.mobile.android | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

Using Windows to make Android smoother Andrews <andrews@spam.net> - 2024-10-23 21:54 +0000
  Re: Using Windows to make Android smoother Paul <nospam@needed.invalid> - 2024-10-24 03:20 -0400
    Re: Using Windows to make Android smoother "Carlos E.R." <robin_listas@es.invalid> - 2024-10-24 16:12 +0200
      Re: Using Windows to make Android smoother "...winston" <winstonmvp@gmail.com> - 2024-10-24 13:20 -0400
        Re: Using Windows to make Android smoother Harry S Robins <stanleyrobins@nothere.uk> - 2024-10-24 13:00 -0500
          Re: Using Windows to make Android smoother "...winston" <winstonmvp@gmail.com> - 2024-10-24 14:12 -0400
          Re: Using Windows to make Android smoother Arno Welzel <usenet@arnowelzel.de> - 2024-10-26 12:10 +0200
            Re: Using Windows to make Android smoother "Carlos E.R." <robin_listas@es.invalid> - 2024-10-26 15:56 +0200
            Re: Using Windows to make Android smoother Paul <nospam@needed.invalid> - 2024-10-26 14:09 -0400
              Re: Using Windows to make Android smoother Arno Welzel <usenet@arnowelzel.de> - 2024-10-28 16:53 +0100
                Re: Using Windows to make Android smoother "Carlos E.R." <robin_listas@es.invalid> - 2024-11-01 09:19 +0100
        Re: Using Windows to make Android smoother Paul <nospam@needed.invalid> - 2024-10-24 14:33 -0400
    Re: Using Windows to make Android smoother Andrews <andrews@spam.net> - 2024-10-24 17:15 +0000
      Re: Using Windows to make Android smoother Paul <nospam@needed.invalid> - 2024-10-24 14:42 -0400
  Re: Using Windows to make Android smoother Andrews <andrews@spam.net> - 2024-10-24 17:35 +0000
    Re: Using Windows to make Android smoother "...winston" <winstonmvp@gmail.com> - 2024-10-24 14:35 -0400
  Re: Using Windows to make Android smoother Andrews <andrews@spam.net> - 2024-10-25 18:41 +0000
  Re: Using Windows to make Android smoother Andrews <andrews@spam.net> - 2024-11-01 06:26 +0000
    Re: Using Windows to make Android smoother Andrews <andrews@spam.net> - 2024-11-01 13:56 +0000

csiph-web