2.0 -//Pentabarf//Schedule//EN PUBLISH MHWKXJ@@conf.gnu-tools-cauldron.org -MHWKXJ Demand Bits, SCEV Assumptions and Other Fun. en en 20261002T090000 20261002T100000 010000

Demand Bits, SCEV Assumptions and Other Fun.

PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/MHWKXJ/ Refectory (S 120) Robin Dapp PUBLISH LCFMB9@@conf.gnu-tools-cauldron.org -LCFMB9 glibc malloc experiments and future direction en en 20261002T103000 20261002T113000 010000

glibc malloc experiments and future direction

We would like to encourage attendees to share their own results and observations. PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/LCFMB9/ Refectory (S 120) Florian Weimer PUBLISH CNN8WR@@conf.gnu-tools-cauldron.org -CNN8WR C++ Support in GCC Static Analyzer: Progress and Future Plans en en 20261002T113000 20261002T120000 003000

C++ Support in GCC Static Analyzer: Progress and Future Plans

This talk is a mix between covering my GSoC project and sharing it with the community, and talking about the state of the project in general. PUBLIC CONFIRMED Talk (Short) https://conf.gnu-tools-cauldron.org/prg26/talk/CNN8WR/ Refectory (S 120) Egas Ribeiro PUBLISH ACPKEV@@conf.gnu-tools-cauldron.org -ACPKEV Compact Unwind Information for ELF Platforms en en 20261002T120000 20261002T123000 003000

Compact Unwind Information for ELF Platforms

Unwind information is included in almost all executables shared libraries for multiple purposes: in-process stack unwinding for, e.g., C++ exceptions and Rust panics, unwinding in debuggers, and stack tracing, e.g. for profiling or crash information. For the non-exception use cases, unwind information must not only be correct at call sites but needs be accurate at every instruction (asynchronous unwind info). In ELF, unwind information is typically encoded in two sections: .eh_frame, which contains one entry (FDE) per function describing the stack frame layout as a DWARF bytecode program, and .eh_frame_hdr, which contains a binary search table of all functions and their entry in .eh_frame. However, these sections often contribute a considerable size overhead to binaries, typically around 3-7%. Further, the stack frame layout of the functions is often somewhat similar, resulting in a comparably redundant encoding. Additionally, the complexity of the DWARF bytecode adds a substantial cost to stack unwinding and prevents an effective use in time-constrained situations like perf stack tracing. In this talk, we will present a compact unwind information format that substantially reduces the unwind info size while also greatly simplifying stack unwinding and tracing through a more simple format without bytecode. Our current implementation in ld.lld supports x86-64 and can accurately represent almost all (>99%) functions generated by GCC and Clang -O3 (on libLLVM.so) (unsupported functions fall back to DWARF bytecode). This compact representation reduces the size of unwind information (.eh_frame_hdr+.eh_frame) by 71% (GCC -O3)/87% (Clang -O3) without loss of information. In addition to size savings, this format also reduces the cost and worst-case runtime of stack unwinding/tracing and can serve as foundation towards low-overhead profiling with fully accurate call traces, without the performance loss and inaccuracies caused by frame pointers. PUBLIC CONFIRMED Talk (Short) https://conf.gnu-tools-cauldron.org/prg26/talk/ACPKEV/ Refectory (S 120) Alexis Engelke PUBLISH KDPJHR@@conf.gnu-tools-cauldron.org -KDPJHR GCC Rust update en en 20261002T133000 20261002T141500 004500

GCC Rust update

PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/KDPJHR/ Refectory (S 120) Pierre-Emmanuel Patry Enes Cevik Janet Chien PUBLISH VB9NV7@@conf.gnu-tools-cauldron.org -VB9NV7 A GPU-Centric View of DWARF 6 en en 20261002T141500 20261002T150000 004500

A GPU-Centric View of DWARF 6

PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/VB9NV7/ Refectory (S 120) Baris Aktemur PUBLISH K93GXZ@@conf.gnu-tools-cauldron.org -K93GXZ Source-Tracking Breakpoints for GDB en en 20261002T150000 20261002T153000 003000

Source-Tracking Breakpoints for GDB

I wrote an article about this feature: https://fedoramagazine.org/gdb-source-tracking-breakpoints/ PUBLIC CONFIRMED Talk (Short) https://conf.gnu-tools-cauldron.org/prg26/talk/K93GXZ/ Refectory (S 120) alexandra petlanova PUBLISH A8QUJX@@conf.gnu-tools-cauldron.org -A8QUJX Steering Committee Q&A en en 20261002T160000 20261002T170000 010000

Steering Committee Q&A

PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/A8QUJX/ Refectory (S 120) David Edelsohn PUBLISH XAHX9K@@conf.gnu-tools-cauldron.org -XAHX9K Practical Formalization of Programming Language Semantics en en 20261002T140000 20261002T153000 013000

Practical Formalization of Programming Language Semantics

PUBLIC CONFIRMED Tutorial https://conf.gnu-tools-cauldron.org/prg26/talk/XAHX9K/ Room S 9 Jose Marchesi PUBLISH YZQYFQ@@conf.gnu-tools-cauldron.org -YZQYFQ The Sourceware Forge Experiment - year 2 and onwards en en 20261002T140000 20261002T143000 003000

The Sourceware Forge Experiment - year 2 and onwards

PUBLIC CONFIRMED Talk (Short) https://conf.gnu-tools-cauldron.org/prg26/talk/YZQYFQ/ Room S 5 Richard Earnshaw PUBLISH MSU8HA@@conf.gnu-tools-cauldron.org -MSU8HA Quantifying Register Pressure: 16-Bit Thumb Constraints on Standard Optimization Passes en en 20261002T143000 20261002T150000 003000

Quantifying Register Pressure: 16-Bit Thumb Constraints on Standard Optimization Passes

I will present using a simple C source my analysis on LICM for target Cortex-M0 plus. I will present results comparing Embench built for size and speed with and without this change. PUBLIC CONFIRMED Birds of a Feather (BoF) https://conf.gnu-tools-cauldron.org/prg26/talk/MSU8HA/ Room S 5 Visda Vokhshoori PUBLISH BFJEP7@@conf.gnu-tools-cauldron.org -BFJEP7 Reverse engineering cost models for the GCC instruction scheduler: a case study on the BananaPi BPI F3 RISC-V CPU en en 20261002T150000 20261002T153000 003000

Reverse engineering cost models for the GCC instruction scheduler: a case study on the BananaPi BPI F3 RISC-V CPU

We attempted to devise tests which give a better understanding of how the HW would respond to changes in the scheduler cost model. This means we don’t pick scheduler costs for instructions randomly, rather, initial values are guided by the measurements observed. We showcase the fine-tuning concepts on the scalar and vector cost model of Spacemit-X60 along with an overview of the GCC internal scheduling cost model implementation and how it works. Measurements presented are obtained using the perf Linux C API. We showcase the challenges of using the perf C API to obtain truthful insights into the inner workings of a RISC-V Linux CPU and how by cross-referencing different methods for evaluating micro-architectural performance, sensible values for a GCC scheduler tuning model can be found. PUBLIC CONFIRMED Talk (Short) https://conf.gnu-tools-cauldron.org/prg26/talk/BFJEP7/ Room S 5 Dusan Stojkovic PUBLISH QJ8X9L@@conf.gnu-tools-cauldron.org -QJ8X9L Introducing new afdo-tools for AutoFDO in GCC en en 20261003T090000 20261003T093000 003000

Introducing new afdo-tools for AutoFDO in GCC

PUBLIC CONFIRMED Talk (Short) https://conf.gnu-tools-cauldron.org/prg26/talk/QJ8X9L/ Room S 9 Kugan Vivekanandarajah PUBLISH YVMRNE@@conf.gnu-tools-cauldron.org -YVMRNE Improving autofdo with perf-gcov en en 20261003T093000 20261003T100000 003000

Improving autofdo with perf-gcov

PUBLIC CONFIRMED Talk (Short) https://conf.gnu-tools-cauldron.org/prg26/talk/YVMRNE/ Room S 9 Andi Kleen PUBLISH BM3WHC@@conf.gnu-tools-cauldron.org -BM3WHC Configury/Build System BoF en en 20261003T103000 20261003T113000 010000

Configury/Build System BoF

PUBLIC CONFIRMED Birds of a Feather (BoF) https://conf.gnu-tools-cauldron.org/prg26/talk/BM3WHC/ Room S 9 pietro PUBLISH HS9RPC@@conf.gnu-tools-cauldron.org -HS9RPC Smart Alignment - Reducing Code Size on x86 without Performance Impact en en 20261003T113000 20261003T120000 003000

Smart Alignment - Reducing Code Size on x86 without Performance Impact

PUBLIC CONFIRMED Talk (Short) https://conf.gnu-tools-cauldron.org/prg26/talk/HS9RPC/ Room S 9 Sarvesh Chandra Venkataramanan Kumar Branden Moore PUBLISH S8AGLY@@conf.gnu-tools-cauldron.org -S8AGLY Debugging of C/C++ Macros like Functions en en 20261003T120000 20261003T121500 001500

Debugging of C/C++ Macros like Functions

Additionally, I will present a demo of the developed [plugin](https://github.com/cast-tech/GDB-Macro-Debugger). PUBLIC CONFIRMED Lightning Talks https://conf.gnu-tools-cauldron.org/prg26/talk/S8AGLY/ Room S 9 Hayk Aslanyan PUBLISH BXDCZA@@conf.gnu-tools-cauldron.org -BXDCZA Beyond Branches: Arm PMU/SPE as a Memory-Side Profile Source for GCC AutoFDO en en 20261003T121500 20261003T123000 001500

Beyond Branches: Arm PMU/SPE as a Memory-Side Profile Source for GCC AutoFDO

This talk is mainly for: people who work on profile/FDO infrastructure, those driving the AutoFDO tooling, and anyone whose workloads are stalled on memory latency rather than branch prediction. In this mechanism, the profile side extends the AutoFDO reader beyond execution counts: cache misses and other PMU events each get their own file, with every count indexed by source location, so an optimisation pass can look up the count for any location. The consumer pass works in six steps. It picks memory-intensive loops using two thresholds, the share and the count of memory instructions; traces each data reference back to its base variable, including indirect and vectorised references; extrapolates data ranges and loop bounds across the nest; filters the loops again by static heuristics or hot paths from the profile; groups references by base variable and ranks them by data volume and access cost; and finally emits prefetch hints for the top groups, naming the target cache level, with both contiguous and SVE gather forms. Where the data volume is computable at compile time and fits the last-level cache, the prefetch is emitted directly; otherwise a runtime check guards it. The memory-intensity thresholds, the target cache level, how many groups to take, and the per-core last-level-cache budget are all tunable parameters. PUBLIC CONFIRMED Lightning Talks https://conf.gnu-tools-cauldron.org/prg26/talk/BXDCZA/ Room S 9 huangxiaoquan PUBLISH Y83C3C@@conf.gnu-tools-cauldron.org -Y83C3C GDB on Windows: status & plans (2026 edition) en en 20261003T133000 20261003T143000 010000

GDB on Windows: status & plans (2026 edition)

PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/Y83C3C/ Room S 9 Pedro Alves PUBLISH PWJ87E@@conf.gnu-tools-cauldron.org -PWJ87E Data-type profiling: pairing perf with pahole en en 20261003T143000 20261003T153000 010000

Data-type profiling: pairing perf with pahole

PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/PWJ87E/ Room S 9 Arnaldo Carvalho de Melo PUBLISH RDQK7U@@conf.gnu-tools-cauldron.org -RDQK7U The GDB BoF en en 20261003T160000 20261003T170000 010000

The GDB BoF

PUBLIC CONFIRMED Birds of a Feather (BoF) https://conf.gnu-tools-cauldron.org/prg26/talk/RDQK7U/ Room S 9 Pedro Alves PUBLISH 3HRHA9@@conf.gnu-tools-cauldron.org -3HRHA9 Register Allocator and Shrink Wrapping en en 20261003T090000 20261003T094500 004500

Register Allocator and Shrink Wrapping

PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/3HRHA9/ Room S 5 Surya Kumari J PUBLISH ARPWWW@@conf.gnu-tools-cauldron.org -ARPWWW Helping Released GCCs Support New CPUs: From Field Problems to a Cross-Version Interface Proposal en en 20261003T094500 20261003T100000 001500

Helping Released GCCs Support New CPUs: From Field Problems to a Cross-Version Interface Proposal

PUBLIC CONFIRMED Lightning Talks https://conf.gnu-tools-cauldron.org/prg26/talk/ARPWWW/ Room S 5 huangxiaoquan PUBLISH L9WJP7@@conf.gnu-tools-cauldron.org -L9WJP7 Manual vectorization for GNU Fortran en en 20261003T103000 20261003T110000 003000

Manual vectorization for GNU Fortran

PUBLIC CONFIRMED Talk (Short) https://conf.gnu-tools-cauldron.org/prg26/talk/L9WJP7/ Room S 5 Magnus Weinmüller Henri Menke PUBLISH C3PSEQ@@conf.gnu-tools-cauldron.org -C3PSEQ Offloading, OpenMP, and OpenACC in GCC en en 20261003T110000 20261003T120000 010000

Offloading, OpenMP, and OpenACC in GCC

PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/C3PSEQ/ Room S 5 Tobias Burnus Thomas Schwinge PUBLISH CDZHQJ@@conf.gnu-tools-cauldron.org -CDZHQJ libgOMPT: Enabling OMPT for libgomp en en 20261003T120000 20261003T123000 003000

libgOMPT: Enabling OMPT for libgomp

Join us for a talk on ongoing work to add OpenMP Tools Interface (OMPT) support to libgomp. OMPT is key for performance analysis and tooling, but until now GCC-compiled OpenMP programs have had limited access to the OMPT-based tooling ecosystem. We will present the new core OMPT infrastructure in libgomp, including callback handling, API entry points, and device callback support. We will also discuss performance results using the EPCC benchmark suite and outline proposed architectural changes for upstream libgomp. This talk should be interesting for anyone working on OpenMP runtimes, GCC/libgomp, performance tools, or the future of tooling support in the GNU ecosystem. PUBLIC CONFIRMED Talk (Short) https://conf.gnu-tools-cauldron.org/prg26/talk/CDZHQJ/ Room S 5 Jan Kraus Thomas Schwinge PUBLISH TXRXTF@@conf.gnu-tools-cauldron.org -TXRXTF Enabling glibc Shadow Stack Runtime in Fedora en en 20261003T133000 20261003T140000 003000

Enabling glibc Shadow Stack Runtime in Fedora

PUBLIC CONFIRMED Talk (Short) https://conf.gnu-tools-cauldron.org/prg26/talk/TXRXTF/ Room S 5 Arjun Shankar PUBLISH 9S88YF@@conf.gnu-tools-cauldron.org -9S88YF BoF on GCC and AI en en 20261003T140000 20261003T150000 010000

BoF on GCC and AI

The goal is to bring together people actively working at this intersection (or interested in it) to compare notes on what's worked, and leave with a clearer sense of concrete next steps. Some possible discussion topics: - What's changed since last year: new accelerator targets, upstream activity etc. - Experiences building GCC/toolchain support for custom NPU and vector extensions: what's reusable versus target-specific. - Where is external tooling (runtime-level) a better fit? PUBLIC CONFIRMED Birds of a Feather (BoF) https://conf.gnu-tools-cauldron.org/prg26/talk/9S88YF/ Room S 5 Pietra Ferreira PUBLISH YF3MFA@@conf.gnu-tools-cauldron.org -YF3MFA BPF BoF en en 20261003T150000 20261003T153000 003000

BPF BoF

PUBLIC CONFIRMED Birds of a Feather (BoF) https://conf.gnu-tools-cauldron.org/prg26/talk/YF3MFA/ Room S 5 David Faust Vineet Gupta Jose Marchesi PUBLISH 7R9YLV@@conf.gnu-tools-cauldron.org -7R9YLV What's new with diagnostics in GCC 17 en en 20261003T160000 20261003T170000 010000

What's new with diagnostics in GCC 17

PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/7R9YLV/ Room S 5 David Malcolm PUBLISH TZERQV@@conf.gnu-tools-cauldron.org -TZERQV GNU C Library BoF en en 20261003T090000 20261003T100000 010000

GNU C Library BoF

This BoF aims to bring together developers of other components that have dependencies on glibc and glibc developers to talk about the following topics: * Planning for Core Toolchain Infrastructure transition and open questions. * Planning for glibc 2.45 and what work needs to be done between September 2026 and February 2027. * Planning for glibc 2.46 and what work needs to be done between February 2027 and August 2027. ... and more. PUBLIC CONFIRMED Birds of a Feather (BoF) https://conf.gnu-tools-cauldron.org/prg26/talk/TZERQV/ Room S (Floor 3) Carlos O'Donell Adhemerval Zanella Netto PUBLISH L7UHWJ@@conf.gnu-tools-cauldron.org -L7UHWJ x86 BoF en en 20261003T103000 20261003T113000 010000

x86 BoF

Discussions are warmly welcomed in the BoF. Here is what we will prepare in advance for background: For APX part, we will have a "brief" introduction for all the features. For ACE part, we will mainly focus on the current implementation in GCC17, and our implementation difference with LLVM. For codesize investigation, we have noticed that, comparing to GCC13, GCC16 has a big increasement on codesize though we also gained a lot for performance. We will share what we are doing and going to do to control the inflation of codesize. PUBLIC CONFIRMED Birds of a Feather (BoF) https://conf.gnu-tools-cauldron.org/prg26/talk/L7UHWJ/ Room S (Floor 3) Haochen Jiang Venkataramanan Kumar Hongtao Liu PUBLISH ALWGLG@@conf.gnu-tools-cauldron.org -ALWGLG Towards real-time ABI compatibility assurance with libctf en en 20261003T113000 20261003T120000 003000

Towards real-time ABI compatibility assurance with libctf

Design overview of a scheme to use type information in C programs and shared libraries to provide (fast, linear-in-number-of-symbols) nearly-precise reporting on broken ABIs, both in libabigail and in ld.so itself (yes, it's that fast). I'll be focusing on interaction with glibc and libabigail mostly so that people can tell me if the whole idea breaks as soon as it hits programs in the real world (and if it's even desirable to do this at all). PUBLIC CONFIRMED Talk (Short) https://conf.gnu-tools-cauldron.org/prg26/talk/ALWGLG/ Room S (Floor 3) Nick Alcock PUBLISH 3PPGKX@@conf.gnu-tools-cauldron.org -3PPGKX Lazy Dynamic Linking for Linux: Deferring Relative Relocations with eBPF en en 20261003T120000 20261003T123000 003000

Lazy Dynamic Linking for Linux: Deferring Relative Relocations with eBPF

PUBLIC CONFIRMED Talk (Short) https://conf.gnu-tools-cauldron.org/prg26/talk/3PPGKX/ Room S (Floor 3) Tal Zussman PUBLISH 9PMLD9@@conf.gnu-tools-cauldron.org -9PMLD9 Arm & AArch64 BoF en en 20261003T133000 20261003T143000 010000

Arm & AArch64 BoF

PUBLIC CONFIRMED Birds of a Feather (BoF) https://conf.gnu-tools-cauldron.org/prg26/talk/9PMLD9/ Room S (Floor 3) Andre Vieira Tamar Christina PUBLISH AVZDSN@@conf.gnu-tools-cauldron.org -AVZDSN Vectorizer BOF en en 20261003T143000 20261003T153000 010000

Vectorizer BOF

PUBLIC CONFIRMED Birds of a Feather (BoF) https://conf.gnu-tools-cauldron.org/prg26/talk/AVZDSN/ Room S (Floor 3) Richard Biener Tamar Christina Robin Dapp PUBLISH 9QEWWH@@conf.gnu-tools-cauldron.org -9QEWWH RISC-V BOF en en 20261003T160000 20261003T170000 010000

RISC-V BOF

PUBLIC CONFIRMED Birds of a Feather (BoF) https://conf.gnu-tools-cauldron.org/prg26/talk/9QEWWH/ Room S (Floor 3) Jeff Law Robin Dapp PUBLISH DLBVLV@@conf.gnu-tools-cauldron.org -DLBVLV GNU Algol 68 BoF en en 20261004T090000 20261004T100000 010000

GNU Algol 68 BoF

PUBLIC CONFIRMED Birds of a Feather (BoF) https://conf.gnu-tools-cauldron.org/prg26/talk/DLBVLV/ Room S 9 Jose Marchesi pietro Mohammad-Reza Nabipoor PUBLISH FXG777@@conf.gnu-tools-cauldron.org -FXG777 KCFI Comes to GCC: Code-Reuse Defenses for the Linux Kernel en en 20261004T103000 20261004T113000 010000

KCFI Comes to GCC: Code-Reuse Defenses for the Linux Kernel

PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/FXG777/ Room S 9 Kees Cook PUBLISH 98FKUB@@conf.gnu-tools-cauldron.org -98FKUB Speeding up ABI analysis of big ELF/DWARF binaries with multi-threading en en 20261004T113000 20261004T123000 010000

Speeding up ABI analysis of big ELF/DWARF binaries with multi-threading

PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/98FKUB/ Room S 9 Dodji Seketeli PUBLISH 9ULSQA@@conf.gnu-tools-cauldron.org -9ULSQA GPU programming with GUTA: GCC's Unified Threading Abstraction en en 20261004T133000 20261004T140000 003000

GPU programming with GUTA: GCC's Unified Threading Abstraction

PUBLIC CONFIRMED Talk (Short) https://conf.gnu-tools-cauldron.org/prg26/talk/9ULSQA/ Room S 9 Thomas Schwinge PUBLISH TRZN7Z@@conf.gnu-tools-cauldron.org -TRZN7Z Toolchain and Linux Kernel BoF en en 20261004T140000 20261004T150000 010000

Toolchain and Linux Kernel BoF

PUBLIC CONFIRMED Birds of a Feather (BoF) https://conf.gnu-tools-cauldron.org/prg26/talk/TRZN7Z/ Room S 9 Jose Marchesi Miguel Ojeda Alexei Alejandro Colomar Paul McKenney Nick Desaulniers PUBLISH 8CDLKZ@@conf.gnu-tools-cauldron.org -8CDLKZ Practical IPA points-to with Steensgaard's algorithm? Why, how and what's the status en en 20261004T090000 20261004T100000 010000

Practical IPA points-to with Steensgaard's algorithm? Why, how and what's the status

PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/8CDLKZ/ Room S 5 Filip Kastl PUBLISH 7LR8AS@@conf.gnu-tools-cauldron.org -7LR8AS MESA: the middle-end semantic assembler en en 20261004T103000 20261004T113000 010000

MESA: the middle-end semantic assembler

We call MESA a "semantic assembler" because it's an assembler for GENERIC semantics. Each MESA function (with couple exceptions) corresponds 1:1 to a GENERIC function or macro. MESA thereby presents the developer with tractable malleable syntax for precise manipulation of the GENERIC tree. When the front-end completes its work and submits it to the middle-end for compilation, MESA produces a dump of the entire GENERIC tree. The syntax used is Scheme. Each GENERIC node is represented as a Scheme variable or function. Functions are Scheme bindings of GENERIC operations -- functions and macros -- exported to Guile. The MESA dump is "sweetened" by Scheme macros to make those operations less tedious to read. To facilitate human comprehension, node names are taken from constituents in the input language. To prove the dump is correct, the user may use MESA to send it to the compiler. We invoke GNU Guile to read the dump and interpret it as Scheme. MESA itself does no interpretation: the input is Scheme, interpreted by Guile. As the interpreter invokes MESA functions, the middle-end is alerted in just the same way, and same sequence, as if invoked from C. The middle-end never sees an S-expression, and the Guile compiler is unconcerned that some Scheme functions happen to be implemented in C. MESA represents a new way to use and think about compiler technology. Historically of course compilers did not have a middle-end. Of modern compilers, none describes anything similar to MESA. The LLVM IR, for example, represents an idealized assembler. The intermediate representations of Rust and Swift are not interchangeable; each reflects constructs of the original language. MESA by contrast is a Scheme abstraction of the GCC middle-end "machine". Although it originated in the COBOL front end, there is no COBOL in it. It is, one might say, as GENERIC as they come. PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/7LR8AS/ Room S 5 James K. Lowden PUBLISH QAQXDB@@conf.gnu-tools-cauldron.org -QAQXDB Toolchain Licensing BoF en en 20261004T113000 20261004T123000 010000

Toolchain Licensing BoF

PUBLIC CONFIRMED Birds of a Feather (BoF) https://conf.gnu-tools-cauldron.org/prg26/talk/QAQXDB/ Room S 5 Denver Gingerich Ian Kelling PUBLISH SVVBRS@@conf.gnu-tools-cauldron.org -SVVBRS Sourceware Infrastructure Organization and Automation en en 20261004T133000 20261004T143000 010000

Sourceware Infrastructure Organization and Automation

- 20 minutes Overview of available Sourceware resources, automating your CI processes (buildbots, try-bots, pre/post-commit-CI/CD, snapshots, forgejo runners, patchwork's curator, etc.) - 20 minutes introduction to the Sourceware organization and goals for next year - 20 minutes Q&A with Sourceware PLC, FSF and SFC staff https://builder.sourceware.org https://sourceware.org/financials.html https://sourceware.org/sourceware-wiki/servers-and-services-2026/ PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/SVVBRS/ Room S 5 Mark J. Wielaard Denver Gingerich Ian Kelling PUBLISH TPLVAA@@conf.gnu-tools-cauldron.org -TPLVAA Using LLMs in patch generation, patch review and code analysis en en 20261004T090000 20261004T100000 010000

Using LLMs in patch generation, patch review and code analysis

PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/TPLVAA/ Room S (Floor 3) Jeff Law PUBLISH XTNMU3@@conf.gnu-tools-cauldron.org -XTNMU3 Working Groups in GCC en en 20261004T103000 20261004T113000 010000

Working Groups in GCC

PUBLIC CONFIRMED Birds of a Feather (BoF) https://conf.gnu-tools-cauldron.org/prg26/talk/XTNMU3/ Room S (Floor 3) Sudi Jason Merrill Carlos O'Donell David Edelsohn PUBLISH KKJKZK@@conf.gnu-tools-cauldron.org -KKJKZK Converting AMD GCN to SIMT vectorization en en 20261004T113000 20261004T123000 010000

Converting AMD GCN to SIMT vectorization

PUBLIC CONFIRMED Talk https://conf.gnu-tools-cauldron.org/prg26/talk/KKJKZK/ Room S (Floor 3) Andrew Stubbs PUBLISH 3HWNAL@@conf.gnu-tools-cauldron.org -3HWNAL Elfutils: The Road to Thread Safety en en 20261004T133000 20261004T140000 003000

Elfutils: The Road to Thread Safety

PUBLIC CONFIRMED Talk (Short) https://conf.gnu-tools-cauldron.org/prg26/talk/3HWNAL/ Room S (Floor 3) Aaron Merey