On 29 February 2016 at 17:53, Leif Lindholm <[email protected]> wrote: > On Mon, Feb 29, 2016 at 04:36:00PM +0800, Haojian Zhuang wrote: >> 在 02/27/2016 04:42 AM, Leif Lindholm 写道: >> >On Fri, Feb 26, 2016 at 05:34:50PM +0800, Haojian Zhuang wrote: >> >>Since there's percentage calcution, multiply on 32bit variable >> >>will cause overflow. So fix the variables as 64bit. >> > >> >So, this is not technically what it does - it changes the variables to >> >be native integer size, meaning this does not fix anything on AArch32. >> > >> >Given that the existing code already populates one of these by calling >> >AsciiStrHexToUint64, it would seem the correct fix would be to follow >> >the commit message and actually change them to UINT64. >> >> mNumDataBytes = AsciiStrHexToUnit64 (NumBytesString); >> Although AsciiStrHexToUnit64 () is used, the value is still stored as 32-bit >> long. Since mNumDataBytes is declared as UINT32. >> >> I don't agree that it doesn't fix on AArch32. UINT64 is also supported >> in AArch32. For example, unsigned long long int is 64-bit long on AArch32, >> and we could do 64-bit calculation on AArch32. > > Yes, UINT64 is also supported on AArch32 - but the patch turns the > variables into UINTN, not UINT64. UINTN is 32-bit on AArch32. >
Oh, yes. I'll fix it with a second version. Thanks for clarify it. Regards Haojian _______________________________________________ edk2-devel mailing list [email protected] https://lists.01.org/mailman/listinfo/edk2-devel
