Hi all, I have interest in some ideas, some I've seen in the Open project pages (copied straight from there) and some are by my own, they are: 1) Implement MachOWriter and ELFWriter to allow LLVM-based compilers to bypass an external assembler. This may include the idea of an assembler for inline assembly 2) Write a disassembler for machine code that would use TableGen to output MachineInstrs for transformations, optimizations, etc. I read two interesting papers on this so far: * Automated Synthesis Of Efficient Binary Decoders for Retargetable Software Toolkits, DAC 2003 * Generating decision trees for decoding binaries 3) Improve ARM be adding necessary ARMv7 instructions and support for VFPv3 and Thumb-2, all this for the ARMv7 Cortex A8 (I can get a i.MX515 board to play if this project is selected) 4) Improve MIPS be to support the Malta board (I'm working on this already, but I rather not work only with MIPS to change things a little bit, but I'd not bother working on this too) I'll improve the proposals in the GSoC app, but for now I'd just like to hear what the community has to say about the ideas (and what they need the most). Any suggestions are welcome :) btw, I don't mind writing more than one proposal and let the mentors get a decision if they like one :) Thanks -- Bruno Cardoso Lopes http://www.brunocardoso.cc
I'm fairly new to the community, but I'm willing to help out mentoring on the first 2 proposals. I've got a 'vested interest' in both. On Fri, Mar 27, 2009 at 7:59 PM, Bruno Cardoso Lopes <bruno.cardoso at gmail.com> wrote:> Hi all, > > I have interest in some ideas, some I've seen in the Open project > pages (copied straight from there) and some > are by my own, they are: > > 1) Implement MachOWriter and ELFWriter to allow LLVM-based compilers > to bypass an external assembler. > This may include the idea of an assembler for inline assembly > > 2) Write a disassembler for machine code that would use TableGen to > output MachineInstrs for transformations, optimizations, etc. > I read two interesting papers on this so far: > * Automated Synthesis Of Efficient Binary Decoders for Retargetable > Software Toolkits, DAC 2003 > * Generating decision trees for decoding binaries > > 3) Improve ARM be adding necessary ARMv7 instructions and support for > VFPv3 and Thumb-2, all this for the ARMv7 Cortex A8 (I can get a > i.MX515 board to play if this project is selected) > > 4) Improve MIPS be to support the Malta board (I'm working on this > already, but I rather not work only with MIPS to change things a > little bit, but I'd not bother working on this too) > > I'll improve the proposals in the GSoC app, but for now I'd just like > to hear what the community > has to say about the ideas (and what they need the most). Any > suggestions are welcome :) > btw, I don't mind writing more than one proposal and let the mentors > get a decision if they like one :) > > Thanks > > -- > Bruno Cardoso Lopes > http://www.brunocardoso.cc > _______________________________________________ > LLVM Developers mailing list > LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev >
On Mar 27, 2009, at 10:59 AM, Bruno Cardoso Lopes wrote:> Hi all, > > I have interest in some ideas, some I've seen in the Open project > pages (copied straight from there) and some > are by my own, they are: > > 1) Implement MachOWriter and ELFWriter to allow LLVM-based compilers > to bypass an external assembler. > This may include the idea of an assembler for inline assemblyThis is be of tremendous value to the llvm community. Evan> > > 2) Write a disassembler for machine code that would use TableGen to > output MachineInstrs for transformations, optimizations, etc. > I read two interesting papers on this so far: > * Automated Synthesis Of Efficient Binary Decoders for Retargetable > Software Toolkits, DAC 2003 > * Generating decision trees for decoding binaries > > 3) Improve ARM be adding necessary ARMv7 instructions and support for > VFPv3 and Thumb-2, all this for the ARMv7 Cortex A8 (I can get a > i.MX515 board to play if this project is selected) > > 4) Improve MIPS be to support the Malta board (I'm working on this > already, but I rather not work only with MIPS to change things a > little bit, but I'd not bother working on this too) > > I'll improve the proposals in the GSoC app, but for now I'd just like > to hear what the community > has to say about the ideas (and what they need the most). Any > suggestions are welcome :) > btw, I don't mind writing more than one proposal and let the mentors > get a decision if they like one :) > > Thanks > > -- > Bruno Cardoso Lopes > http://www.brunocardoso.cc > _______________________________________________ > LLVM Developers mailing list > LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
On Mar 27, 2009, at 10:59 AM, Bruno Cardoso Lopes wrote:> I have interest in some ideas, some I've seen in the Open project > pages (copied straight from there) and some > are by my own, they are:Hi Bruno,> 1) Implement MachOWriter and ELFWriter to allow LLVM-based compilers > to bypass an external assembler. > This may include the idea of an assembler for inline assembly > > 2) Write a disassembler for machine code that would use TableGen to > output MachineInstrs for transformations, optimizations, etc. > I read two interesting papers on this so far: > * Automated Synthesis Of Efficient Binary Decoders for Retargetable > Software Toolkits, DAC 2003 > * Generating decision trees for decoding binaries > > 3) Improve ARM be adding necessary ARMv7 instructions and support for > VFPv3 and Thumb-2, all this for the ARMv7 Cortex A8 (I can get a > i.MX515 board to play if this project is selected) > > 4) Improve MIPS be to support the Malta board (I'm working on this > already, but I rather not work only with MIPS to change things a > little bit, but I'd not bother working on this too) > > I'll improve the proposals in the GSoC app, but for now I'd just like > to hear what the community > has to say about the ideas (and what they need the most). Any > suggestions are welcome :) > btw, I don't mind writing more than one proposal and let the mentors > get a decision if they like one :)Just MHO, but of the three, I think the first one is the most directly useful in the short term. For either #3/4, it would be very useful to pick a specific device that is widely available, or target a community. I believe that last year, you were targeting PSP development, for example, which aims directly at supporting a specific and concrete community of potential users. Submitting multiple proposals is also perfectly acceptable, and I do hope you apply, -Chris