* Required to support Debian Live 9.1 in ISO mode
* Note that this only works if the efi.img boot files do not require
additional content besides the one extracted from the ISO.
* Add a WaitForSingleObjectWithMessages() call so that we can process Windows messages
while waiting on events (prevents lockup while issuing log messages)
* Limit the total duration of CheckDriveAccess() to 2 seconds
* Allow for user cancellation
* Also update code to use the Edit_####() predefined macros for Edit controls instead of EM_### messages
* The process search appears to be blocking on some platform, and we
also don't want users to have to wait too long on format startup
* Also update the update check for Windows XP SSL errors
* Rufus now checks for processes with handles opened on the drives/volumes before
starting the format operation and asks the user if they want to continue.
* This mimics Windows' behaviour when formatting drives, and actually uses the
same message as the one from shell32.dll.mui.
* Closes#773
* Fix memory leaks due to not releasing the heap
* Also speed up lookups by not reopening the same process twice in a row
* Also make the usb_debug variable global so we can use uuprintf everywhere
* Enable the use of GPT on REMOVABLE drives, if the host is Windows 10 CU
* Don't display the message about REMOVABLE boot, if the target is Windows 10 CU
* Not having FILE_SHARE_WRITE on a disk or volume handle helps
prevent spurious accesses from the OS or other apps while we
are repartitioning/refomatting.
* It appears that 1.5TB and 2TB HDDs, accessed trough some Seagate ow WD USB ↔ SATA
controllers, can report that 0 bytes were written on WriteFile(), even though all
the data was effectively written. 1TB HDDs, accessed through the same controller,
do not report this issue. So add a workaround for that.
* Also see #787
* Such as WebConverger (webc-35.1.iso)
* Also fix non installation of Syslinux if Alt-E (dual UEFI/BIOS) is set
* Also update README and AppVeyor script
* Tooltips can only be manipulated from the same thread as the one that created their parent
* Also increase the SetLGP() timeout as this seems to be problematic
* Closes#764
* Downloadable content will now be indicating, in the log, whether it can be trusted with ✓ (validated) or ✗ (caution)
* Of course this validation only applies for files we know of, i.e. the downloadable content that existed at the time the DB was
created. So, if Syslinux 8.x gets released tomorrow and we put it on our server, you'll get an ✗ regardless of its integrity.
* Closes#758
* SetChecksumAffinity() now spreads the affinity evenly between cores
* Also increase the read buffer size to help with performance
* At this stage, the only limiting factor regarding performance seems
to be the speed of the SHA-256 algorithm...
Pete Batard