9/7/12 Housekeeping ò Welcome TA Amit Arya – Office Hours posted Interrupts and System ò Next Thursday’s class has a reading assignment ò Lab 1 due Friday Calls ò All students should have VMs at this point Don Porter ò Email Don if you don’t have one CSE 506 ò Private git repositories should be set-up Logical Diagram Background: Control Flow Binary Memory Threads pc // x = 2, y = true void printf(va_args) Formats Allocators { User if (y) { //... System Calls Kernel 2 /= x; } RCU File System Networking Sync printf(x); Today’s Lecture } //... Memory Device CPU Management Drivers Scheduler Regular control flow: branches and calls Hardware (logically follows source code) Interrupts Disk Net Consistency Background: Control Flow Lecture goal // x = 0, y = true void handle_divzero() ò Understand the hardware tools available for irregular pc Divide by zero! Program can’t make { control flow. if (y) { progress! x = 2; ò I.e., things other than a branch in a running program 2 /= x; ò Building blocks for context switching, device } printf(x); management, etc. } //... Irregular control flow: exceptions, system calls, etc. 1 9/7/12 Two types of interrupts Intel nomenclature ò Synchronous: will happen every time an instruction ò Interrupt – only refers to asynchronous interrupts executes (with a given program state) ò Exception – synchronous control transfer ò Divide by zero ò System call ò Bad pointer dereference ò Note: from the programmer’s perspective, these are ò Asynchronous: caused by an external event handled with the same abstractions ò Usually device I/O ò Timer ticks (well, clocks can be considered a device) Lecture outline Interrupt overview ò Overview ò Each interrupt or exception includes a number indicating its type ò How interrupts work in hardware ò E.g., 14 is a page fault, 3 is a debug breakpoint ò How interrupt handlers work in software ò This number is the index into an interrupt table ò How system calls work ò New system call hardware on x86 x86 interrupt table x86 interrupt overview Device IRQs 48 = JOS 128 = Linux ò Each type of interrupt is assigned an index from 0—255. System Call System Call ò 0—31 are for processor interrupts; generally fixed by Intel ò E.g., 14 is always for page faults … … … ò 32—255 are software configured 0 31 47 255 ò 32—47 are for device interrupts (IRQs) in JOS ò Most device’s IRQ line can be configured ò Look up APICs for more info (Ch 4 of Bovet and Cesati) ò 0x80 issues system call in Linux (more on this later) Software Configurable Reserved for the CPU 2 9/7/12 Software interrupts Software interrupts, cont ò The int <num> instruction allows software to raise an ò OS sets ring level required to raise an interrupt interrupt ò Generally, user programs can’t issue an int 14 (page ò 0x80 is just a Linux convention. JOS uses 0x30. fault manually) ò There are a lot of spare indices ò An unauthorized int instruction causes a general protection fault ò You could have multiple system call tables for different ò Interrupt 13 purposes or types of processes! ò Windows does: one for the kernel and one for win32k What happens (generally): How it works (HW) ò Control jumps to the kernel ò How does HW know what to execute? ò At a prescribed address (the interrupt handler) ò Where does the HW dump the registers; what does it use ò The register state of the program is dumped on the kernel’s as the interrupt handler’s stack? stack ò Sometimes, extra info is loaded into CPU registers ò E.g., page faults store the address that caused the fault in the cr2 register ò Kernel code runs and handles the interrupt ò When handler completes, resume program (see iret instr.) How is this configured? x86 interrupt table idtr ò Kernel creates an array of Interrupt descriptors in memory, called Interrupt Descriptor Table, or IDT … … … ò Can be anywhere in physical memory ò Pointed to by special register (idtr) 0 31 47 255 ò c.f., segment registers and gdtr and ldtr! ò Entry 0 configures interrupt 0, and so on Physical Address of Interrupt Table (Avoids going through page translation) 3 9/7/12 x86 interrupt table Interrupt Descriptor idtr ò Code segment selector ò Almost always the same (kernel code segment) … … … ò Recall, this was designed before paging on x86! ò Segment offset of the code to run 0 31 47 255 14 ò Kernel segment is “flat”, so this is just the linear address ò Privilege Level (ring) Code Segment: Kernel Code ò Interrupts can be sent directly to user code. Why? Segment Offset: &page_fault_handler //linear addr ò Present bit – disable unused interrupts Ring: 0 // kernel Present: 1 ò Gate type (interrupt or trap/exception) – more in a bit Gate Type: Exception x86 interrupt table Interrupt Descriptors, ctd. idtr ò In-memory layout is a bit confusing ò Like a lot of the x86 architecture, many interfaces were … … … later deprecated 0 3 31 47 255 ò Worth comparing Ch 9.5 of the i386 manual with inc/ mmu.h in the JOS source code Code Segment: Kernel Code Segment Offset: &breakpoint_handler //linear addr Ring: 3 // user Present: 1 Gate Type: Exception How it works (HW) Task State Segment (TSS) ò How does HW know what to execute? ò Another segment, just like the code and data segment ò Interrupt descriptor table specifies what code to run and at ò A descriptor created in the GDT (cannot be in LDT) what privilege ò Selected by special task register (tr) ò This can be set up once during boot for the whole system ò Unlike others, has a hardware-specified layout ò Where does the HW dump the registers; what does it use ò Lots of fields for rarely-used features as the interrupt handler’s stack? ò Two features we care about in a modern OS: ò Specified in the Task State Segment ò 1) Location of kernel stack (fields ss0/esp0) ò 2) I/O Port privileges (more in a later lecture) 4 9/7/12 TSS, cont. Summary ò Simple model: specify a TSS for each process ò Most interrupt handling hardware state set during boot ò Optimization (JOS): ò Each interrupt has an IDT entry specifying: ò Our kernel is pretty simple (uniprocessor only) ò What code to execute, privilege level to raise the interrupt ò Why not just share one TSS and kernel stack per-process? ò Stack to use specified in the TSS ò Linux generalization: ò One TSS per CPU ò Modify TSS fields as part of context switching Comment Lecture outline ò Again, segmentation rears its head ò Overview ò You can’t program OS-level code on x86 without getting ò How interrupts work in hardware your hands dirty with it ò How interrupt handlers work in software ò Helps to know which features are important when reading the manuals ò How system calls work ò New system call hardware on x86 High-level goal Interrupt Handlers ò Respond to some event, return control to the appropriate ò Just plain old kernel code process ò What to do on: ò Network packet arrives ò Disk read completion ò Divide by zero ò System call 5 9/7/12 Example Complication: Stack Stack ò What happens if I’m in an interrupt handler, and another interrupt comes in? RSP Disk RSP Interrupt! ò Note: kernel stack only changes on privilege level change ò Nested interrupts just push the next frame on the stack ò What could go wrong? if (x) { RIP RIP ò Violate code invariants printf(“Boo”); Disk_handler (){ ... ò Deadlock ... } ò Exhaust the stack (if too many fire at once) printf(va_args…){ ... User Kernel Example Bottom Line: Stack Stack ò Interrupt service routines must be reentrant or synchronize Network RSP Interrupt! ò Period. Will Hang Forever!RIP disk_handler (){ lock_kernel(); if (x) { Already Locked!!! ... printf(“Boo”); unlock_kernel(); ... ... printf(va_args…){ net_handler (){ ... lock_kernel(); User Kernel … Hardware interrupt sync. Disabling interrupts ò While a CPU is servicing an interrupt on a given IRQ ò An x86 CPU can disable I/O interrupts line, the same IRQ won’t raise another interrupt until the routine completes ò Clear bit 9 of the EFLAGS register (IF Flag) ò Bottom-line: device interrupt handler doesn’t have to ò cli and sti instructions clear and set this flag worry about being interrupted by itself ò Before touching a shared data structure (or grabbing a ò A different device can interrupt the handler lock), an interrupt handler should disable I/O interrupts ò Problematic if they share data structures ò Like a list of free physical pages… ò What if both try to grab a lock for the free list? 6 9/7/12 Gate types Exceptions ò Recall: an IDT entry can be an interrupt or an exception ò You can’t mask exceptions gate ò Why not? ò Difference? ò Can’t make progress after a divide-by-zero ò An interrupt gate automatically disables all other ò Double and Triple faults detect faults in the kernel interrupts (i.e., clears and sets IF on enter/exit) ò Do exception handlers need to be reentrant? ò An exception gate doesn’t ò Not if your kernel has no bugs (or system calls in itself) ò This is just a programmer convenience: you could do the ò In certain cases, Linux allows nested page faults same thing in software ò E.g., to detect errors copying user-provided buffers Summary Lecture outline ò Interrupt handlers need to synchronize, both with locks ò Overview (multi-processor) and by disabling interrupts (same CPU) ò How interrupts work in hardware ò Exception handlers can’t be masked ò How interrupt handlers work in software ò Nested exceptions generally avoided ò How system calls work ò New system call hardware on x86 System call “interrupt” Lecture outline ò Originally, system calls issued using int instruction ò Overview ò Dispatch routine was just an interrupt handler ò How interrupts work in hardware ò Like interrupts,