For those interested:

http://www.electronicdesign.com/embedded-revolution/xip-linux-new-spin-embedded-architecture

 

Execute-in-Place. Is that where you see a Linux system and put a bullet through its heart?

 

Execute-in-Place. Is that where you see a Linux system and put a bullet through its heart?

No.

 

For those interested:

http://www.electronicdesign.com/embedded-revolution/xip-linux-new-spin-embedded-architecture

Seems to be forcing a square peg into a round hole. There are plenty of embedded applications that actually require a Linux type OS.

 

Seems to be forcing a square peg into a round hole. There are plenty of embedded applications that actually require a Linux type OS.

I think you missed the point of the article.

The kernel runs out of SPI flash -- not DRAM -- with apparently little performance impact.

This is a major improvement in both system cost and boot speed.

 

I think you missed the point of the article.

The kernel runs out of SPI flash -- not DRAM -- with apparently little performance impact.

This is a major improvement in both system cost and boot speed.

It's been possible to run Linux from external flash for a while, they've make it better and faster by making it smaller (stripping to the bare essentials) with a specific file-system designed for XIP Linux. That's great but as the article says: XIP Linux achieved its intended purpose at the time.
https://lwn.net/Articles/135472/

Looks a good sell point for the Renesas RZ/A1 MPU but not a real advance for embedded Linux as a whole. I'm all for Linux but I see mainly a marketing ploy.

 

It's been possible to run Linux from external flash for a while, they've make it better and faster by making it smaller (stripping to the bare essentials) with a specific file-system designed for XIP Linux. That's great but as the article says: XIP Linux achieved its intended purpose at the time.
https://lwn.net/Articles/135472/

Looks a good sell point for the Renesas RZ/A1 MPU but not a real advance for embedded Linux as a whole. I'm all for Linux but I see mainly a marketing ploy.

Have to agree - it's very much a niche.

There have been linux (and before that other *nix) versions that have been able to execute in place. It's not that hard to do. The "disruptive" part is building SPI interface into the instruction fetch of the CPU. My guess is that they have built a cache mechanism with SPI support. I'm skeptical of the "no performance penalty" claim though. They just used one test (file transfer) which doesn't really stress the CPU and shines with more buffer space. I'm pretty sure tests which require less RAM would tell a different tale.