Hello! Here is simple draft of possible text for bug in linux, *bsd kernels.
"Project flashrom (http://www.flashrom.org) trying to get full and free/open-source solution for BIOS reflashing. It works nice on usual PC systems, and becoming new standard of industry. But on laptops, which have one independend processor core - Embedded Conroller (EC) there is a big trouble to get it working safe, because EC usually always interacting with OS over ACPI. And if flashrom try to flash BIOS, when some of ACPI interrupt calls, it can make laptop dead. " Best regards, Anton Kochkov.
to put this into some perspective... we (mainly twice11, xvilka, me) were discussing "the" major architectural problem of flashrom's internal programmer on irc: doing delicate stuff in user space without kernel support.
we came to the conclusion that it is worth to at least try to convince kernel people to provide interfaces for us to get more control over important things like acpi' execution and its interrupts.
On Sat, 23 Apr 2011, Stefan Tauner wrote:
to put this into some perspective... we (mainly twice11, xvilka, me) were discussing "the" major architectural problem of flashrom's internal programmer on irc: doing delicate stuff in user space without kernel support.
we came to the conclusion that it is worth to at least try to convince kernel people to provide interfaces for us to get more control over important things like acpi' execution and its interrupts.
I doubt if kernel people of any such distro allow userspace program to get control about acpi, but maybe they could be motivated for exposing interface which allow to issue some operation where kernel actually talk to underlying hardware but also prevents acpi or other stuff from interrupting.
Of course that would require each kernel to implement similar/overlaping set of features so abstract layer could be done so main part of flashrom code would not need know what kind of underlying kernel interface is needed. Maybe that could be done as new type internal programmer, like -p linux26 which ask kernel for proxying operations, and -p bsd44 and so on so main part and current raw access mode could remain untouched while new interface could be implemented and tested.
best regards Maciej
* Stefan Tauner stefan.tauner@student.tuwien.ac.at [110423 15:53]:
to put this into some perspective... we (mainly twice11, xvilka, me) were discussing "the" major architectural problem of flashrom's internal programmer on irc: doing delicate stuff in user space without kernel support.
we came to the conclusion that it is worth to at least try to convince kernel people to provide interfaces for us to get more control over important things like acpi' execution and its interrupts.
I think the way to do this is to provide a kernel patch that implements such interfaces and then discuss with the kernel guys.
Stefan
-
Maciej Pijanka -
Stefan Reinauer -
Stefan Tauner -
Антон Кочков