On 01/29/2014 04:25 PM, Rafael Espíndola wrote:>> How would this pass get added to the chain? >> >> It has to create more functions and then those in turn need to be compiled. > I am sure it can be done. MergeFunctions for example creates new functions. > > Now, it does seem this request/code review is a bit upside down. You > have been told that your original proposal is not acceptable. > Reviewers may offer to help, but in the end it is the author that has > to go figure out how something has to be done. You seem to expect that > Eric and I are now the ones responsible for it. > > Cheers, > RafaelYou may have missed this in the discussion, but this cannot be done in an IR pass. I find out that I need the stub during ISEL lowering. Reed
Which part of lowering requires this? -eric On Wed, Jan 29, 2014 at 4:42 PM, reed kotler <rkotler at mips.com> wrote:> On 01/29/2014 04:25 PM, Rafael Espíndola wrote: >>> >>> How would this pass get added to the chain? >>> >>> It has to create more functions and then those in turn need to be >>> compiled. >> >> I am sure it can be done. MergeFunctions for example creates new >> functions. >> >> Now, it does seem this request/code review is a bit upside down. You >> have been told that your original proposal is not acceptable. >> Reviewers may offer to help, but in the end it is the author that has >> to go figure out how something has to be done. You seem to expect that >> Eric and I are now the ones responsible for it. >> >> Cheers, >> Rafael > > You may have missed this in the discussion, but this cannot be done in an IR > pass. > > I find out that I need the stub during ISEL lowering. > > Reed >
On 01/29/2014 04:53 PM, Eric Christopher wrote:> Which part of lowering requires this? > > -ericIf I see that certain externals are referenced, then I know I need to generate a stub for them. Most of this I take care of in an IR pass but some I can't know about because they are just standard libc type functions. There are only a few that create this problem for mips16 hard float when it is linked as c++. I intend though to move all the earlier actual machine code generation from the IR pass to this same place that I am using to solve the problem at hand.> On Wed, Jan 29, 2014 at 4:42 PM, reed kotler <rkotler at mips.com> wrote: >> On 01/29/2014 04:25 PM, Rafael Espíndola wrote: >>>> How would this pass get added to the chain? >>>> >>>> It has to create more functions and then those in turn need to be >>>> compiled. >>> I am sure it can be done. MergeFunctions for example creates new >>> functions. >>> >>> Now, it does seem this request/code review is a bit upside down. You >>> have been told that your original proposal is not acceptable. >>> Reviewers may offer to help, but in the end it is the author that has >>> to go figure out how something has to be done. You seem to expect that >>> Eric and I are now the ones responsible for it. >>> >>> Cheers, >>> Rafael >> You may have missed this in the discussion, but this cannot be done in an IR >> pass. >> >> I find out that I need the stub during ISEL lowering. >> >> Reed >>