Common stuff for all my devices' web server pages: css, login, log, ipv4, ipv6, firmware update, clock, reset info etc.
A password has to be set whenever there has been a software reset. Resets following faults or power on do not require a new password as the hash is restored from the RTC GPREG register.
The password is not saved on the device; instead a 32 bit hash of the password is saved. It would take 2^31 attempts to brute force the password: this could be done in under a month if an attempt were possible every millisecond. To prevent this a 200 ms delay is introduced in the reply to the login form, that gives a more reasonable 13 years to brute force the password.
Once the password is accepted a random session id is created. This is 36 bit to give six base 64 characters but without an extra delay. If an attempt could be made every ms then this would still take over a year to brute force.
The most likely attack would to use a dictionary with, say, 10 million entries against the password which would still take 20 days to do.
Revisions of web.c
|125:772948168e4f||6 months ago||Updated net library||File Diff Annotate|
|123:06de83222fda||6 months ago||Updated http module in the net library||File Diff Annotate|
|122:cd3f391ac8aa||6 months ago||Updated http||File Diff Annotate|
|112:f29bb9b99059||6 months ago||Changed all names from 'derived' to 'this'||File Diff Annotate|
|111:aaa858678e34||6 months ago||Corrected bug where postComplete was not set true in the event of there not being a post.||File Diff Annotate|
|110:8ab752842d25||6 months ago||Tidied. About to rename to web.||File Diff Annotate|
|109:3e82f62c7e1f||6 months ago||Tidied names from http to web||File Diff Annotate|