Ok,
I 've just tried and lastly I had the file in the attachment 
1100 seems like a lot of hot pixels, how was the shot taken and what commands did you use in chkdptp?
Generally, to fix hot pixels you would want to
1) Take a "dark frame" with the longest exposure and highest ISO you'd expect to use.
You can make the dark frame either by covering the lens, or using script to keep the shutter closed during the exposure. You could also just use a scene that's very under-exposed, but you will want to use a very conservative limit for what you call "hot" in that case.
Note that if you use an exposure over a few seconds, the Canon firmware will automatically shoot and subtract a dark frame. The camera will show "busy" and the shot will take something like 2x the selected exposure time. The dark frame takes care of hot pixels, so you wouldn't generally want to use an exposure this long to generate a badpixel list. You can disable the cameras automatic dark frame the CHDK raw settings, but in this case you'd usually want to apply your own dark frames later.
2) Load the resulting DNG into chdkptp with dngload file.dng and generate a badpixel.txt containing pixels with value above some limit.
The exact value is matter of taste, but it should be something substantially above the black level, because there will be some amp glow and dark current. If you covered the lens, there will probably be some stray light too. You don't want to try to "fix" all those pixels with the badpixel list, you only want the ones that are really "hot" or stuck on.
You can use the dnghist command to get an idea of where the hot pixels are. Here's an example using an 4 second, ISO800 dark frame from my D10 (12 bit sensor, like your a2500)
$ chdkptp -i
___> dngload CRW_1060.DNG
loaded CRW_1060.DNG
___> dnghist -bin=512
0-511 12338145
512-1023 3
1024-1535 1
1536-2047 3
2048-2559 5
2560-3071 0
3072-3583 0
3584-4095 3
Here we see virtually all the pixels fall below 512. It's probably safe to call the remainder "hot", but we can take a close look at the low end
___> dnghist -bin=32 -max=512
0-31 144385
32-63 607806
64-95 1663674
96-127 2718160
128-159 3720824
160-191 2302111
192-223 857943
224-255 250329
256-287 55785
288-319 12795
320-351 3233
352-383 832
384-415 208
416-447 43
448-479 14
480-511 3
512-543 0
The black level on 12 bit cameras is around 128, so we expect to see the values clustered around there. 4 seconds is quite a long exposure, so we expect some dark current etc too. It's clear that values over 512 are exceptional, so we can use that as the maximum:
___> dnglistpixels -min=512 -fmt=chdk -out=badpixel.txt
15 matching pixels
If the "dark frame" wasn't shot with the shutter closed, there might be a lot of real values around 512, so you would use a higher limit.
Sorry for the many silly questions but..........to make this file working right how should I use it?
Should I put it inside the chdk directory?
Shoud I rename it?
Copy it to the CHDK directory as BADPIXEL or BADPIXEL.TXT
Will it works for all raw format (CR2, CRW) or only for DNG?
If the "Manual bad pxiel removal" setting is not OFF, it will be applied to all images. How it actually works depends on which of the settings you use, see
http://chdk.wikia.com/wiki/CHDK_User_Manual#Manual_bad_pixel_removalFor DNG 1.1, the RawConv option will leave you with 0 value pixels in the data, which will look like unfixed badpixels.
If I choose dng 1.1 the camera will use badpixel.bin instead of the file generated with chdkptp?
No, DNG badpixel.bin does a different thing. See
http://chdk.wikia.com/wiki/Badpixel_removalbadpixel lets you specify pixels as bad. This is generally used for "hot" pixels that are not known to the canon firmware.
badpixel.bin fixes patches pixels the camera sets 0 (ones the canon firmware already knows are bad) in order to make a valid 1.1 DNG.
