I second this request. GPT is in my opinion superior to the MSDOS partition-table type, for several reasons like robustness (no fragile linked list of extended partitions, backup partition table), flexibility and of course scalability, the reason why Mr. DeRobertis wanted it.
As he also points out, it works and is useful on normal ia32 hardware (in spite of what fs/partitions/Kconfig says). I must admit I cannot comment on the potential issues with broken USB devices, as I don't have such hardware in my possession. It does strike me as odd, though, that there is a problem: As I understand it the GPT detection code fails if it doesn't find a valid "Protective MBR", ie. a MSDOS partition table with one partition, spanning the whole drive, with the type set to 0xEE. Am I wrong assuming that such a partition table will almost certainly never be found on such a broken USB device that doesn't have a GPT, and thus there should be no problem? See also the documentation for the 'gpt' kernel parameter in Documentation/kernel-parameters.txt. Finally I'd like to point out that SuSE do indeed support GPT in their ESÂ 9 product, so there must obviously be some way to enable it without USB breakage ensuing. [1] Enterprise Schmenterprise..? ;-) Thanks, -- Tore Anderson
