Dynamo

Bala, Vasanth; Duesterwald, Evelyn; Banerjia, Sanjeev

doi:10.1145/349299.349303

“…MAMBO-X64 uses a variant of the Next Executing Tail (NET) scheme [13] from Dynamo [3] to generate traces. NET works by adding an execution counter to basic blocks which are the target of a backwards branch or an exit from an existing trace, which is incremented every time the basic block is executed.…”

Section: Return-aware Trace Generationmentioning

confidence: 99%

Low overhead dynamic binary translation on ARM

d'Antras

¹

,

Gorgovan

²

,

Garside

³

et al. 2017

Proceedings of the 38th ACM SIGPLAN Conference on Programming Language Design and Implementation

View full text Add to dashboard Cite

The ARMv8 architecture introduced AArch64, a 64-bit execution mode with a new instruction set, while retaining binary compatibility with previous versions of the ARM architecture through AArch32, a 32-bit execution mode. Most hardware implementations of ARMv8 processors support both AArch32 and AArch64, which comes at a cost in hardware complexity.We present MAMBO-X64, a dynamic binary translator for Linux which executes 32-bit ARM binaries using only the AArch64 instruction set. We have evaluated the performance of MAMBO-X64 on three existing ARMv8 processors which support both AArch32 and AArch64 instruction sets. The performance was measured by comparing the running time of 32-bit benchmarks running under MAMBO-X64 with the same benchmark running natively. On SPEC CPU2006, we achieve a geometric mean overhead of less than 7.5 % on in-order Cortex-A53 processors and a performance improvement of 1 % on out-of-order X-Gene 1 processors.MAMBO-X64 achieves such low overhead by novel optimizations to map AArch32 floating-point registers to AArch64 registers dynamically, handle overflowing address calculations efficiently, generate traces that harness hardware return address prediction, and handle operating system signals accurately.

show abstract

“…Load and store instructions in both AArch32 and AArch64 support a similar set of addressing modes 3 , of which a few examples are shown in Table 3. Despite their similarity however, simply translating the AArch32 addressing modes into their AArch64 equivalent will not always produce correct results.…”

Section: Speculative Address Generationmentioning

confidence: 99%

“…• Previous research in Dynamo [3] has shown that effective trace generation in a DBT can improve runtime performance compared to native execution.…”

Section: Overall Performancementioning

confidence: 99%

Low overhead dynamic binary translation on ARM

D'AntrasAmanieu

¹

,

GorgovanCosmin

²

,

GarsideJim

³

et al. 2017

SIGPLAN Not.

1

0

View full text Add to dashboard Cite

The ARMv8 architecture introduced AArch64, a 64-bit execution mode with a new instruction set, while retaining binary compatibility with previous versions of the ARM architecture through AArch32, a 32-bit execution mode. Most hardware implementations of ARMv8 processors support both AArch32 and AArch64, which comes at a cost in hardware complexity.We present MAMBO-X64, a dynamic binary translator for Linux which executes 32-bit ARM binaries using only the AArch64 instruction set. We have evaluated the performance of MAMBO-X64 on three existing ARMv8 processors which support both AArch32 and AArch64 instruction sets. The performance was measured by comparing the running time of 32-bit benchmarks running under MAMBO-X64 with the same benchmark running natively. On SPEC CPU2006, we achieve a geometric mean overhead of less than 7.5 % on in-order Cortex-A53 processors and a performance improvement of 1 % on out-of-order X-Gene 1 processors.MAMBO-X64 achieves such low overhead by novel optimizations to map AArch32 floating-point registers to AArch64 registers dynamically, handle overflowing address calculations efficiently, generate traces that harness hardware return address prediction, and handle operating system signals accurately.

show abstract

“…Machine code to machine code JIT translation has been used to speed up execution in the Dynamo system (Bala et al, 2000). It is not used to insert additional monitoring code into the generated code.…”

Section: Related Workmentioning

confidence: 99%

Protecting Private Data in the Cloud

Rasmusson¹,

Aslam²

2012

Proceedings of the 2nd International Conference on Cloud Computing and Services Science

0

View full text Add to dashboard Cite

Abstract:Companies that process business critical and secret data are reluctant to use utility and cloud computing for the risk that their data gets stolen by rogue system administrators at the hosting company. We describe a system organization that prevents host administrators from directly accessing or installing eaves-dropping software on the machine that holds the client's valuable data. Clients are monitored via machine code probes that are inlined into the clients' programs at runtime. The system enables the cloud provider to install and remove software probes into the machine code without stopping the client's program, and it prevents the provider from installing probes not granted by the client.

show abstract

Dynamo

Cited by 519 publications

References 20 publications

Low overhead dynamic binary translation on ARM

Low overhead dynamic binary translation on ARM

Low overhead dynamic binary translation on ARM

Protecting Private Data in the Cloud

Contact Info

Product

Resources

About