Understanding modern device drivers

Kadav, Asim; Swift, Michael M.

doi:10.1145/2150976.2150987

Cited by 47 publications

(16 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Cloning in the Linux kernel has been extensively studied in the literature [15][16] [17][21] [24] mainly focusing on the detection of small cloned code fragments. In the Linux v2.4.0, Casazza et al [17] reported cloning of 15.5% between arch and drivers subsystems.…”

Section: A Cloning In the Linux Kernelmentioning

confidence: 99%

See 1 more Smart Citation

Managing Big Clones to Ease Evolution: Linux Kernel Example

Kumar

Jarzabek

Dan³

2016

Annals of Computer Science and Information Systems

View full text Add to dashboard Cite

Abstract-Successful software is often enhanced and adapted to the needs of new users. During evolution, a software system grows in size, becomes more complex, and costly to maintain. In this paper, we point to big clones-large granular duplicated program structures such as files or directories-as one of many reasons why this happens. Using the Linux kernel as an example, we show that big clones arise in the Linux kernel despite careful architecture design and a systematic approach for managing variability. We propose a solution to avoid these big clones by representing them as generalized templates in ART (Adaptive Reuse Technique). ART templates are constructed on top of the Linux code, without conflicts with the state-of-art techniques and tools used to manage the Linux kernel. Benefits include simplification of the Linux kernel due to non-redundancy, easier comprehension, and traceability of the change impact during evolution. The proposed technique is general and the Linux example discussed in this paper also illustrates general phenomena.

show abstract

Section: A Cloning In the Linux Kernelmentioning

confidence: 99%

“…Other study showed that file subsystem had 12% clone coverage in the Linux v2.4.19 [16]. In the Linux v2.6.37.6, 8% code similarity between drivers (/sound and /drivers directories) has been reported [15]. An empirical study of cloning among SCSI drivers is done by [24].…”

Section: A Cloning In the Linux Kernelmentioning

confidence: 99%

Managing Big Clones to Ease Evolution: Linux Kernel Example

Kumar

Jarzabek

Dan³

2016

Annals of Computer Science and Information Systems

View full text Add to dashboard Cite

show abstract

“…Drivers use kernel extension functions to operate and manage devices. However, in some cases, these functions may return errors that are even fatal to drivers, such as allocating memory faults and configuration faults [1]. And in error handling paths, the driver should release the acquired resources and execute the operations that are opposite to its previous execution.…”

Section: Introductionmentioning

confidence: 99%

PF-Miner: A New Paired Functions Mining Method for Android Kernel in Error Paths

Liu

Wang

Jiang

et al. 2014

2014 IEEE 38th Annual Computer Software and Applications Conference

View full text Add to dashboard Cite

Drivers are significant components of the operating systems(OSs), and they run in kernel mode. Generally, drivers have many errors to handle, and the functions called in the normal execution paths and error handling paths are in pairs, which are named as paired functions. However, some developers do not handle the errors completely as they forget about or are unaware of releasing the acquired resources, thus memory leaks and other potential problems can be easily introduced. Therefore, it is highly valuable to automatically extract paired functions for these problems and detect violations for the programmers.This paper proposes an efficient tool named PFMiner, which can automatically extract paired functions and detect violations between normal execution paths and error handling paths from the source code of C program with the data mining and statistical methods. We have evaluated PFMiner on different versions of Android kernel 2.6.39 and 3.10.0, and 81 bugs reported by PF-Miner in 2.6.39 have been fixed before the latest version 3.10.0. PF-Miner only needs about 150 seconds to analyze the source code of 3.10.0, and 983 violations have been detected from 546 paired functions that have been extracted. We have reported the top 51 violations as potential bugs to the developers, and 15 bugs have been confirmed.

show abstract

“…However, many drivers have proprietary commands that cannot be captured by a shadow driver common to an entire class, leading to incomplete recovery. Recent work showed that up to 44% of drivers have non-class behavior [21].…”

Section: Introductionmentioning

confidence: 99%

Fine-grained fault tolerance using device checkpoints

Kadav

Renzelmann

Swift

2013

Proceedings of the Eighteenth International Conference on Architectural Support for Programming Languages and Operating Systems

Self Cite

View full text Add to dashboard Cite

Recovering faults in drivers is difficult compared to other code because their state is spread across both memory and a device. Existing driver fault-tolerance mechanisms either restart the driver and discard its state, which can break applications, or require an extensive logging mechanism to replay requests and recreate driver state. Even logging may be insufficient, though, if the semantics of requests are ambiguous. In addition, these systems either require large subsystems that must be kept up-to-date as the kernel changes, or require substantial rewriting of drivers. We present a new driver fault-tolerance mechanism that provides fine-grained control over the code protected. Fine-Grained Fault Tolerance (FGFT) isolates driver code at the granularity of a single entry point. It executes driver code as a transaction, allowing roll back if the driver fails. We develop a novel checkpointing mechanism to save and restore device state using existing power-management code. Unlike past systems, FGFT can be incremen-tally deployed in a single driver without the need for a large kernel subsystem, but at the cost of small modifications to the driver. In the evaluation, we show that FGFT can have almost zero run-time cost in many cases, and that checkpoint-based recovery can reduce the duration of a failure by 79% compared to restarting the driver. Finally, we show that applying FGFT to a driver requires little effort, and the majority of drivers in common classes already contain the power-management code needed for check-point/restore.

show abstract

Understanding modern device drivers

Cited by 47 publications

References 26 publications

Managing Big Clones to Ease Evolution: Linux Kernel Example

Managing Big Clones to Ease Evolution: Linux Kernel Example

PF-Miner: A New Paired Functions Mining Method for Android Kernel in Error Paths

Fine-grained fault tolerance using device checkpoints

Contact Info

Product

Resources

About