Differences

This shows you the differences between two versions of the page.

--- etc:users:jcmvbkbc:little-things:4 [2010/01/04 04:02] – plan jcmvbkbc
+++ etc:users:jcmvbkbc:little-things:4 [2016/08/08 20:53] (current) – ↷ Page moved from users:jcmvbkbc:little-things:4 to etc:users:jcmvbkbc:little-things:4 kel
@@ Line 7: / Line 7: @@
 ===== Worklog =====
+  * Modify qemu (easy part): add timer to qemu device, use qemu_irq_pulse inside its callback;
+  * Make PCI device driver: start with init and exit fns, pci_driver structure and device table;
+  * Make PCI probe fn; main steps:
+    * pci_enable_device (enable device IO and memory);
+    * request_irq (still there's no device open fn do it in probe);
+    * pci_set_drvdata (to make our device accessible from the struct device level);
+  * Make PCI release fn, do cleanup of what probe done;
+  * Add simple irq handler and sysfs output;
+Code: ftp://kkv.spb.su/pub/home/dumb/ws/lted/20100104
 ===== Conclusion =====
+  * when to assert/deassert IRQ line: no matter (at least for emulation);
+    * although it is generally level-triggered, qemu treats them as edge-triggered, so that qemu_irq_pulse may be used;
+    * anyway (e.g. for proper IRQ sharing) separate 'irq status' register is needed for the device function (not implemented in this demo, but hold on :);
+  * when to start/stop interrupts: doc contradiction;
+    * PCI spec says that IRQ is enabled (interrupt disable == 0 in command register) upon reset, but linux pci documentation says that IRQ will go only when you explicitly enable them;
+    * anyway, nothing happens (at least, nothing is visible) until request_irq;
+    * even more: right place to request_irq is not probe routine, but rather device open routine;
+  * qemu 0.12.1 yields maximum of ~3000 PCI interrupts/second;
+  * using tabs in Linux is stupid ):
 {{tag>qemu PCI IRQ lted}}