Transitioned from Windows to Linux kernel.
-
After many many years doing Windows kernel with a bit of Linux, perhaps a few months at most, I have transitioned to the Linux kernel on a probably majority basis, given the lack of Windows kernel work these days. And it is interesting. On Windows I wrote mostly USB device drivers recently, on Linux I get to write the USB HCI driver that interrats directly with the hub. This is much more fun, prodding arond in the registers, setig bits here and there, Reminds me of NT4 with its monolythic drivers that sit directly on the hardware. One thign of interest, is just how similar the kernels are, a Linux SoftIrq is a Windows DISPATCH_LEVEL call, spinlocks, semaphores and events are the same, of course the hardware and language are the same, adn the only real difference is the architectural model, and the inter driver communication. IRPs on Windows and Structs with funtion ponters on Linux. Have to say though that Wingdb, the Windows kernel debugger is way way beter than the kgdb/gdb combo on Linux, and the inability to build a non optimised Linux kernel makes stepping through code limited. But its been an easy transition all in all, far easier than I expected. I wouldnt think going from Linux to Windows would be though, the ramp up in complexity is considerable.
-
After many many years doing Windows kernel with a bit of Linux, perhaps a few months at most, I have transitioned to the Linux kernel on a probably majority basis, given the lack of Windows kernel work these days. And it is interesting. On Windows I wrote mostly USB device drivers recently, on Linux I get to write the USB HCI driver that interrats directly with the hub. This is much more fun, prodding arond in the registers, setig bits here and there, Reminds me of NT4 with its monolythic drivers that sit directly on the hardware. One thign of interest, is just how similar the kernels are, a Linux SoftIrq is a Windows DISPATCH_LEVEL call, spinlocks, semaphores and events are the same, of course the hardware and language are the same, adn the only real difference is the architectural model, and the inter driver communication. IRPs on Windows and Structs with funtion ponters on Linux. Have to say though that Wingdb, the Windows kernel debugger is way way beter than the kgdb/gdb combo on Linux, and the inability to build a non optimised Linux kernel makes stepping through code limited. But its been an easy transition all in all, far easier than I expected. I wouldnt think going from Linux to Windows would be though, the ramp up in complexity is considerable.
I have no idea about most of what you're talking about, but it's compelling reading.
I wanna be a eunuchs developer! Pass me a bread knife!
-
After many many years doing Windows kernel with a bit of Linux, perhaps a few months at most, I have transitioned to the Linux kernel on a probably majority basis, given the lack of Windows kernel work these days. And it is interesting. On Windows I wrote mostly USB device drivers recently, on Linux I get to write the USB HCI driver that interrats directly with the hub. This is much more fun, prodding arond in the registers, setig bits here and there, Reminds me of NT4 with its monolythic drivers that sit directly on the hardware. One thign of interest, is just how similar the kernels are, a Linux SoftIrq is a Windows DISPATCH_LEVEL call, spinlocks, semaphores and events are the same, of course the hardware and language are the same, adn the only real difference is the architectural model, and the inter driver communication. IRPs on Windows and Structs with funtion ponters on Linux. Have to say though that Wingdb, the Windows kernel debugger is way way beter than the kgdb/gdb combo on Linux, and the inability to build a non optimised Linux kernel makes stepping through code limited. But its been an easy transition all in all, far easier than I expected. I wouldnt think going from Linux to Windows would be though, the ramp up in complexity is considerable.
I can ignore every other typo, but please spell WinDbg correctly. :)
"Real men drive manual transmission" - Rajesh.
-
After many many years doing Windows kernel with a bit of Linux, perhaps a few months at most, I have transitioned to the Linux kernel on a probably majority basis, given the lack of Windows kernel work these days. And it is interesting. On Windows I wrote mostly USB device drivers recently, on Linux I get to write the USB HCI driver that interrats directly with the hub. This is much more fun, prodding arond in the registers, setig bits here and there, Reminds me of NT4 with its monolythic drivers that sit directly on the hardware. One thign of interest, is just how similar the kernels are, a Linux SoftIrq is a Windows DISPATCH_LEVEL call, spinlocks, semaphores and events are the same, of course the hardware and language are the same, adn the only real difference is the architectural model, and the inter driver communication. IRPs on Windows and Structs with funtion ponters on Linux. Have to say though that Wingdb, the Windows kernel debugger is way way beter than the kgdb/gdb combo on Linux, and the inability to build a non optimised Linux kernel makes stepping through code limited. But its been an easy transition all in all, far easier than I expected. I wouldnt think going from Linux to Windows would be though, the ramp up in complexity is considerable.
-
I did some work on AIX drivers some years back, starting from scratch, and that had really good kernel tracing in it. My only problem was the lack of formal training so I was learning and writing at the same time.
Use the best guess
Richard MacCutchan wrote:
My only problem was the lack of formal training so I was learning and writing at the same time.
Does this have anything to do with your signature? :)
"Real men drive manual transmission" - Rajesh.
-
Richard MacCutchan wrote:
My only problem was the lack of formal training so I was learning and writing at the same time.
Does this have anything to do with your signature? :)
"Real men drive manual transmission" - Rajesh.
-
I have no idea about most of what you're talking about, but it's compelling reading.
I wanna be a eunuchs developer! Pass me a bread knife!
I have got gdb on my mind... oh my god, I have beome a Linuxer!
-
I can ignore every other typo, but please spell WinDbg correctly. :)
"Real men drive manual transmission" - Rajesh.
I have got gdb on my mind... oh my god, I have beome a Linuxer!
-
After many many years doing Windows kernel with a bit of Linux, perhaps a few months at most, I have transitioned to the Linux kernel on a probably majority basis, given the lack of Windows kernel work these days. And it is interesting. On Windows I wrote mostly USB device drivers recently, on Linux I get to write the USB HCI driver that interrats directly with the hub. This is much more fun, prodding arond in the registers, setig bits here and there, Reminds me of NT4 with its monolythic drivers that sit directly on the hardware. One thign of interest, is just how similar the kernels are, a Linux SoftIrq is a Windows DISPATCH_LEVEL call, spinlocks, semaphores and events are the same, of course the hardware and language are the same, adn the only real difference is the architectural model, and the inter driver communication. IRPs on Windows and Structs with funtion ponters on Linux. Have to say though that Wingdb, the Windows kernel debugger is way way beter than the kgdb/gdb combo on Linux, and the inability to build a non optimised Linux kernel makes stepping through code limited. But its been an easy transition all in all, far easier than I expected. I wouldnt think going from Linux to Windows would be though, the ramp up in complexity is considerable.
Given that the head of the Windows Driver Development group is named S. A. Tan, any other kernel development gig has to be an improvement.
Software Zen:
delete this; -
After many many years doing Windows kernel with a bit of Linux, perhaps a few months at most, I have transitioned to the Linux kernel on a probably majority basis, given the lack of Windows kernel work these days. And it is interesting. On Windows I wrote mostly USB device drivers recently, on Linux I get to write the USB HCI driver that interrats directly with the hub. This is much more fun, prodding arond in the registers, setig bits here and there, Reminds me of NT4 with its monolythic drivers that sit directly on the hardware. One thign of interest, is just how similar the kernels are, a Linux SoftIrq is a Windows DISPATCH_LEVEL call, spinlocks, semaphores and events are the same, of course the hardware and language are the same, adn the only real difference is the architectural model, and the inter driver communication. IRPs on Windows and Structs with funtion ponters on Linux. Have to say though that Wingdb, the Windows kernel debugger is way way beter than the kgdb/gdb combo on Linux, and the inability to build a non optimised Linux kernel makes stepping through code limited. But its been an easy transition all in all, far easier than I expected. I wouldnt think going from Linux to Windows would be though, the ramp up in complexity is considerable.
Happy linux user & kernel hacker available here :-D
Beauty cannot be defined by abscissas and ordinates; neither are circles and ellipses created by their geometrical formulas.
-
Given that the head of the Windows Driver Development group is named S. A. Tan, any other kernel development gig has to be an improvement.
Software Zen:
delete this;I know someone called S.A. Tan. His parents didn't think of it because in the Far East, he is Stephen Tan Aun: they put the Christian name first, surname in the middle and given Chinese names at the end. His initials were STA. When he came to the west, he put his given name before the surname and became Stephen Aun Tan or S.A.Tan.
-
After many many years doing Windows kernel with a bit of Linux, perhaps a few months at most, I have transitioned to the Linux kernel on a probably majority basis, given the lack of Windows kernel work these days. And it is interesting. On Windows I wrote mostly USB device drivers recently, on Linux I get to write the USB HCI driver that interrats directly with the hub. This is much more fun, prodding arond in the registers, setig bits here and there, Reminds me of NT4 with its monolythic drivers that sit directly on the hardware. One thign of interest, is just how similar the kernels are, a Linux SoftIrq is a Windows DISPATCH_LEVEL call, spinlocks, semaphores and events are the same, of course the hardware and language are the same, adn the only real difference is the architectural model, and the inter driver communication. IRPs on Windows and Structs with funtion ponters on Linux. Have to say though that Wingdb, the Windows kernel debugger is way way beter than the kgdb/gdb combo on Linux, and the inability to build a non optimised Linux kernel makes stepping through code limited. But its been an easy transition all in all, far easier than I expected. I wouldnt think going from Linux to Windows would be though, the ramp up in complexity is considerable.