***Server sets mode: +nt
<AlmuHS>me voy a dormir (I go to sleep). Adios (bye) ;) <gnu_srs2>(13:48:56) damo22: lets see if this thing works: How did it go? <damo22>__save_flags(x) and __restore_flags(x) in linux drivers might be wrong <damo22>im trying to remove pic_mask and setting of PIC masks <damo22>and instead just letting mach enable and disable interrupts totally <damo22>can someone check my logic here: if i am saving the flags inside a linux driver, i need to test if the current IPL is greater than zero, and if it is, set the linux "flags" variable to (1 << EFLAGS_IF_SHIFT) otherwise zero <damo22>spl0 means interrupts are enabled, spl >0 means interrupts are enabled <damo22>spl0 means interrupts are enabled, spl >0 means interrupts are disabled <damo22>if the state of interrupts is currently disabled, is the IF set or cleared? <youpi>sti means setting the interrupt flag <youpi>so if ipl is greater than zero, the if should be cleared <damo22>also linux expects the flags variable to be raw flag, not 0 or 7 ipl <damo22>i think this __SMP__ path is opening a new can of worms that has not been looked at before because up to this point we never had more than one processor using linux drivers? <damo22>hmm but i am not using the smp code <gnu_srs2>I asked you this a few days ago: (23:08:04) gnu_srs2: damo22: Doesn't combining your work with the smp development unnecessary complicated? <damo22>gnu_srs2: i wanted to attempt adding APIC support and it needs ACPI which almuhs added <damo22>so it was logical to work ontop of his branch <damo22>i have it mostly working there, but i can simplify it a lot if i do what you are saying <damo22>so now i am working on debian kernel as a base <damo22>attempting some simplifications of the irq handling <damo22>i have one patch so far that can probably be added <damo22>and im working on the second patch <damo22>is that the AC config option that gets enabled if we are building with the kernel debugger? <youpi>you can send the first patch already ***Emulatorman__ is now known as Emulatorman
<youpi>and there you have the glibc-2.30 branch ***slyfox_ is now known as slyfox
<gnu_srs2>youpi: Detecting pkg-config, libpciaccess, etc in configure for hurd does not work when cross-compiling. <gnu_srs2>Have to find out which rules to use, any ideas?