On Mon, 2015-08-31 at 03:50 +0300, Siarhei Siamashka wrote: > > > -swaptbl_start: > > > +appended_data: > > > > If I'm reading correctly you've changed things so that there is an > > implicit data structure present here, namely the spl addr followed > by > > the swaptbl (as before). > > That's right. The format of the data has changed because we need to > additionally pass the SPL load address to the thunk code. I had a > hope that with the use of named aliases for the registers, the source > code is already reasonably clear and self-descriptive: > > ldr SPL_ADDR, appended_data > > adr SWAPTBL, appended_data + 4
Right, the named registers help, it was these initialisations which triggered the thought rather than the usage. [...] > Yes, it is not very desirable. Right now the C code is treating the > generated 'fel_to_spl_thunk' array as just a piece of position > independent ARM code. Without trying to make any assumptions about > what's inside of it. And then appends the required data to it. > If we change the C code to do patching of the last word in this > array, then things become a bit more messy on the C side instead > of the ASM side. I suspected as much. > > If it isn't possible/desirable maybe just a comment describing what > > should live here and/or some #define's for the offsets would be > useful? > > A good point. Would something like this be good enough? Absolutely. Ian. -- You received this message because you are subscribed to the Google Groups "linux-sunxi" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
