Searched hist:2010 (Results 551 - 575 of 929) sorted by relevance
/gem5/src/arch/arm/isa/templates/ | ||
H A D | basic.isa | diff 7168:54105b48a0a7 Wed Jun 02 01:58:00 EDT 2010 Ali Saidi <Ali.Saidi@ARM.com> ARM: GCC < 4.3 has some issues with attribute no return on some functions. Fix so it works for older gccs. 7167:a28390624772 Wed Jun 02 01:58:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> ARM: Split out the "basic" templates and format. |
/gem5/src/arch/power/isa/formats/ | ||
H A D | mem.isa | diff 7712:7733c562e5e3 Fri Oct 22 03:23:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> ISA: Simplify various implementations of completeAcc. diff 7045:e21fe6a62b1c Tue Mar 23 11:50:00 EDT 2010 Steve Reinhardt <steve.reinhardt@amd.com> cpu: fix exec tracing memory corruption bug Accessing traceData (to call setAddress() and/or setData()) after initiating a timing translation was causing crashes, since a failed translation could delete the traceData object before returning. It turns out that there was never a need to access traceData after initiating the translation, as the traced data was always available earlier; this ordering was merely historical. Furthermore, traceData->setAddress() and traceData->setData() were being called both from the CPU model and the ISA definition, often redundantly. This patch standardizes all setAddress and setData calls for memory instructions to be in the CPU models and not in the ISA definition. It also moves those calls above the translation calls to eliminate the crashes. |
H A D | branch.isa | diff 7791:762276cd3cc7 Wed Dec 08 13:33:00 EST 2010 Gabe Black <gblack@eecs.umich.edu> POWER: Take advantage of new PCState syntax. diff 7720:65d338a8dba4 Sun Oct 31 03:07:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> ISA,CPU,etc: Create an ISA defined PC type that abstracts out ISA behaviors. This change is a low level and pervasive reorganization of how PCs are managed in M5. Back when Alpha was the only ISA, there were only 2 PCs to worry about, the PC and the NPC, and the lsb of the PC signaled whether or not you were in PAL mode. As other ISAs were added, we had to add an NNPC, micro PC and next micropc, x86 and ARM introduced variable length instruction sets, and ARM started to keep track of mode bits in the PC. Each CPU model handled PCs in its own custom way that needed to be updated individually to handle the new dimensions of variability, or, in the case of ARMs mode-bit-in-the-pc hack, the complexity could be hidden in the ISA at the ISA implementation's expense. Areas like the branch predictor hadn't been updated to handle branch delay slots or micropcs, and it turns out that had introduced a significant (10s of percent) performance bug in SPARC and to a lesser extend MIPS. Rather than perpetuate the problem by reworking O3 again to handle the PC features needed by x86, this change was introduced to rework PC handling in a more modular, transparent, and hopefully efficient way. PC type: Rather than having the superset of all possible elements of PC state declared in each of the CPU models, each ISA defines its own PCState type which has exactly the elements it needs. A cross product of canned PCState classes are defined in the new "generic" ISA directory for ISAs with/without delay slots and microcode. These are either typedef-ed or subclassed by each ISA. To read or write this structure through a *Context, you use the new pcState() accessor which reads or writes depending on whether it has an argument. If you just want the address of the current or next instruction or the current micro PC, you can get those through read-only accessors on either the PCState type or the *Contexts. These are instAddr(), nextInstAddr(), and microPC(). Note the move away from readPC. That name is ambiguous since it's not clear whether or not it should be the actual address to fetch from, or if it should have extra bits in it like the PAL mode bit. Each class is free to define its own functions to get at whatever values it needs however it needs to to be used in ISA specific code. Eventually Alpha's PAL mode bit could be moved out of the PC and into a separate field like ARM. These types can be reset to a particular pc (where npc = pc + sizeof(MachInst), nnpc = npc + sizeof(MachInst), upc = 0, nupc = 1 as appropriate), printed, serialized, and compared. There is a branching() function which encapsulates code in the CPU models that checked if an instruction branched or not. Exactly what that means in the context of branch delay slots which can skip an instruction when not taken is ambiguous, and ideally this function and its uses can be eliminated. PCStates also generally know how to advance themselves in various ways depending on if they point at an instruction, a microop, or the last microop of a macroop. More on that later. Ideally, accessing all the PCs at once when setting them will improve performance of M5 even though more data needs to be moved around. This is because often all the PCs need to be manipulated together, and by getting them all at once you avoid multiple function calls. Also, the PCs of a particular thread will have spatial locality in the cache. Previously they were grouped by element in arrays which spread out accesses. Advancing the PC: The PCs were previously managed entirely by the CPU which had to know about PC semantics, try to figure out which dimension to increment the PC in, what to set NPC/NNPC, etc. These decisions are best left to the ISA in conjunction with the PC type itself. Because most of the information about how to increment the PC (mainly what type of instruction it refers to) is contained in the instruction object, a new advancePC virtual function was added to the StaticInst class. Subclasses provide an implementation that moves around the right element of the PC with a minimal amount of decision making. In ISAs like Alpha, the instructions always simply assign NPC to PC without having to worry about micropcs, nnpcs, etc. The added cost of a virtual function call should be outweighed by not having to figure out as much about what to do with the PCs and mucking around with the extra elements. One drawback of making the StaticInsts advance the PC is that you have to actually have one to advance the PC. This would, superficially, seem to require decoding an instruction before fetch could advance. This is, as far as I can tell, realistic. fetch would advance through memory addresses, not PCs, perhaps predicting new memory addresses using existing ones. More sophisticated decisions about control flow would be made later on, after the instruction was decoded, and handed back to fetch. If branching needs to happen, some amount of decoding needs to happen to see that it's a branch, what the target is, etc. This could get a little more complicated if that gets done by the predecoder, but I'm choosing to ignore that for now. Variable length instructions: To handle variable length instructions in x86 and ARM, the predecoder now takes in the current PC by reference to the getExtMachInst function. It can modify the PC however it needs to (by setting NPC to be the PC + instruction length, for instance). This could be improved since the CPU doesn't know if the PC was modified and always has to write it back. ISA parser: To support the new API, all PC related operand types were removed from the parser and replaced with a PCState type. There are two warts on this implementation. First, as with all the other operand types, the PCState still has to have a valid operand type even though it doesn't use it. Second, using syntax like PCS.npc(target) doesn't work for two reasons, this looks like the syntax for operand type overriding, and the parser can't figure out if you're reading or writing. Instructions that use the PCS operand (which I've consistently called it) need to first read it into a local variable, manipulate it, and then write it back out. Return address stack: The return address stack needed a little extra help because, in the presence of branch delay slots, it has to merge together elements of the return PC and the call PC. To handle that, a buildRetPC utility function was added. There are basically only two versions in all the ISAs, but it didn't seem short enough to put into the generic ISA directory. Also, the branch predictor code in O3 and InOrder were adjusted so that they always store the PC of the actual call instruction in the RAS, not the next PC. If the call instruction is a microop, the next PC refers to the next microop in the same macroop which is probably not desirable. The buildRetPC function advances the PC intelligently to the next macroop (in an ISA specific way) so that that case works. Change in stats: There were no change in stats except in MIPS and SPARC in the O3 model. MIPS runs in about 9% fewer ticks. SPARC runs with 30%-50% fewer ticks, which could likely be improved further by setting call/return instruction flags and taking advantage of the RAS. TODO: Add != operators to the PCState classes, defined trivially to be !(a==b). Smooth out places where PCs are split apart, passed around, and put back together later. I think this might happen in SPARC's fault code. Add ISA specific constructors that allow setting PC elements without calling a bunch of accessors. Try to eliminate the need for the branching() function. Factor out Alpha's PAL mode pc bit into a separate flag field, and eliminate places where it's blindly masked out or tested in the PC. |
/gem5/src/dev/arm/ | ||
H A D | rv_ctrl.cc | diff 7733:08d6a773d1b6 Mon Nov 08 14:58:00 EST 2010 Ali Saidi <Ali.Saidi@ARM.com> ARM: Add checkpointing support 7584:28ddf6d9e982 Mon Aug 23 12:18:00 EDT 2010 Ali Saidi <Ali.Saidi@arm.com> ARM: Add I/O devices for booting linux |
H A D | pl011.hh | diff 7587:177151a54462 Mon Aug 23 12:18:00 EDT 2010 Ali Saidi <Ali.Saidi@arm.com> ARM: Change how the AMBA device ID checking is done to make it more generic 7584:28ddf6d9e982 Mon Aug 23 12:18:00 EDT 2010 Ali Saidi <Ali.Saidi@arm.com> ARM: Add I/O devices for booting linux |
/gem5/src/mem/slicc/ast/ | ||
H A D | FuncCallExprAST.py | diff 7793:f6cbeb8712d3 Wed Dec 08 14:52:00 EST 2010 Brad Beckmann <Brad.Beckmann@amd.com> ruby: remove Ruby asserts for m5.fast This diff is for changing the way ASSERT is handled in Ruby. m5.fast compiles out the assert statements by using the macro NDEBUG. Ruby uses the macro RUBY_NO_ASSERT to do so. This macro has been removed and NDEBUG has been put in its place. diff 7780:42da07116e12 Wed Dec 01 14:30:00 EST 2010 Nilay Vaish <nilay@cs.wisc.edu> ruby: Converted old ruby debug calls to M5 debug calls This patch developed by Nilay Vaish converts all the old GEMS-style ruby debug calls to the appropriate M5 debug calls. |
H A D | __init__.py | diff 7567:238f99c9f441 Fri Aug 20 14:46:00 EDT 2010 Brad Beckmann <Brad.Beckmann@amd.com> ruby: Stall and wait input messages instead of recycling This patch allows messages to be stalled in their input buffers and wait until a corresponding address changes state. In order to make this work, all in_ports must be ranked in order of dependence and those in_ports that may unblock an address, must wake up the stalled messages. Alot of this complexity is handled in slicc and the specification files simply annotate the in_ports. diff 6882:898047a3672c Fri Jan 29 23:29:00 EST 2010 Brad Beckmann <Brad.Beckmann@amd.com> ruby: Ruby changes required to use the python config system This patch includes the necessary changes to connect ruby objects using the python configuration system. Mainly it consists of removing unnecessary ruby object pointers and connecting the necessary object pointers using the generated param objects. This patch includes the slicc changes necessary to connect generated ruby objects together using the python configuraiton system. |
/gem5/src/arch/alpha/ | ||
H A D | registers.hh | diff 7699:addb847910d2 Mon Oct 04 14:58:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> Alpha: Fix Alpha NumMiscArchRegs constant. Also add asserts in O3's Scoreboard class to catch bad indexes. diff 7649:a6a6177a5ffa Wed Aug 25 20:10:00 EDT 2010 Min Kyu Jeong <minkyu.jeong@arm.com> ARM: Fixed register flattening logic (FP_Base_DepTag was set too low) When decoding a srs instruction, invalid mode encoding returns invalid instruction. This can happen when garbage instructions are fetched from mispredicted path |
/gem5/src/arch/arm/linux/ | ||
H A D | linux.hh | diff 7441:be2acdfb8bdc Wed Jun 02 01:58:00 EDT 2010 Ali Saidi <Ali.Saidi@ARM.com> ARM: Fix SPEC2000 benchmarks in SE mode. With this patch all Spec2k benchmarks seem to run with atomic or timing mode simple CPUs. Fixed up some constants, handling of 64 bit arguments, and marked a few more syscalls ignoreFunc. diff 7416:e1a7a9f33a00 Wed Jun 02 01:58:00 EDT 2010 Ali Saidi <Ali.Saidi@ARM.com> ARM: fix sizes of structs for ARM Linux |
/gem5/src/arch/arm/ | ||
H A D | process.hh | diff 7441:be2acdfb8bdc Wed Jun 02 01:58:00 EDT 2010 Ali Saidi <Ali.Saidi@ARM.com> ARM: Fix SPEC2000 benchmarks in SE mode. With this patch all Spec2k benchmarks seem to run with atomic or timing mode simple CPUs. Fixed up some constants, handling of 64 bit arguments, and marked a few more syscalls ignoreFunc. diff 7096:e81026b9dfe0 Wed Jun 02 01:58:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> ARM: Allow ARM processes to start in Thumb mode. |
H A D | utility.cc | diff 7752:08e1e28a062a Mon Nov 15 15:04:00 EST 2010 William Wang <William.Wang@arm.com> ARM: Add support for GDB on ARM diff 7749:859e8bc1cdc2 Mon Nov 15 15:04:00 EST 2010 Ali Saidi <Ali.Saidi@ARM.com> ARM: Cache the misc regs at the TLB to limit readMiscReg() calls. diff 7748:7bf78d12b359 Mon Nov 15 15:04:00 EST 2010 Ali Saidi <Ali.Saidi@ARM.com> ARM: Add support for switching CPUs diff 7720:65d338a8dba4 Sun Oct 31 03:07:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> ISA,CPU,etc: Create an ISA defined PC type that abstracts out ISA behaviors. This change is a low level and pervasive reorganization of how PCs are managed in M5. Back when Alpha was the only ISA, there were only 2 PCs to worry about, the PC and the NPC, and the lsb of the PC signaled whether or not you were in PAL mode. As other ISAs were added, we had to add an NNPC, micro PC and next micropc, x86 and ARM introduced variable length instruction sets, and ARM started to keep track of mode bits in the PC. Each CPU model handled PCs in its own custom way that needed to be updated individually to handle the new dimensions of variability, or, in the case of ARMs mode-bit-in-the-pc hack, the complexity could be hidden in the ISA at the ISA implementation's expense. Areas like the branch predictor hadn't been updated to handle branch delay slots or micropcs, and it turns out that had introduced a significant (10s of percent) performance bug in SPARC and to a lesser extend MIPS. Rather than perpetuate the problem by reworking O3 again to handle the PC features needed by x86, this change was introduced to rework PC handling in a more modular, transparent, and hopefully efficient way. PC type: Rather than having the superset of all possible elements of PC state declared in each of the CPU models, each ISA defines its own PCState type which has exactly the elements it needs. A cross product of canned PCState classes are defined in the new "generic" ISA directory for ISAs with/without delay slots and microcode. These are either typedef-ed or subclassed by each ISA. To read or write this structure through a *Context, you use the new pcState() accessor which reads or writes depending on whether it has an argument. If you just want the address of the current or next instruction or the current micro PC, you can get those through read-only accessors on either the PCState type or the *Contexts. These are instAddr(), nextInstAddr(), and microPC(). Note the move away from readPC. That name is ambiguous since it's not clear whether or not it should be the actual address to fetch from, or if it should have extra bits in it like the PAL mode bit. Each class is free to define its own functions to get at whatever values it needs however it needs to to be used in ISA specific code. Eventually Alpha's PAL mode bit could be moved out of the PC and into a separate field like ARM. These types can be reset to a particular pc (where npc = pc + sizeof(MachInst), nnpc = npc + sizeof(MachInst), upc = 0, nupc = 1 as appropriate), printed, serialized, and compared. There is a branching() function which encapsulates code in the CPU models that checked if an instruction branched or not. Exactly what that means in the context of branch delay slots which can skip an instruction when not taken is ambiguous, and ideally this function and its uses can be eliminated. PCStates also generally know how to advance themselves in various ways depending on if they point at an instruction, a microop, or the last microop of a macroop. More on that later. Ideally, accessing all the PCs at once when setting them will improve performance of M5 even though more data needs to be moved around. This is because often all the PCs need to be manipulated together, and by getting them all at once you avoid multiple function calls. Also, the PCs of a particular thread will have spatial locality in the cache. Previously they were grouped by element in arrays which spread out accesses. Advancing the PC: The PCs were previously managed entirely by the CPU which had to know about PC semantics, try to figure out which dimension to increment the PC in, what to set NPC/NNPC, etc. These decisions are best left to the ISA in conjunction with the PC type itself. Because most of the information about how to increment the PC (mainly what type of instruction it refers to) is contained in the instruction object, a new advancePC virtual function was added to the StaticInst class. Subclasses provide an implementation that moves around the right element of the PC with a minimal amount of decision making. In ISAs like Alpha, the instructions always simply assign NPC to PC without having to worry about micropcs, nnpcs, etc. The added cost of a virtual function call should be outweighed by not having to figure out as much about what to do with the PCs and mucking around with the extra elements. One drawback of making the StaticInsts advance the PC is that you have to actually have one to advance the PC. This would, superficially, seem to require decoding an instruction before fetch could advance. This is, as far as I can tell, realistic. fetch would advance through memory addresses, not PCs, perhaps predicting new memory addresses using existing ones. More sophisticated decisions about control flow would be made later on, after the instruction was decoded, and handed back to fetch. If branching needs to happen, some amount of decoding needs to happen to see that it's a branch, what the target is, etc. This could get a little more complicated if that gets done by the predecoder, but I'm choosing to ignore that for now. Variable length instructions: To handle variable length instructions in x86 and ARM, the predecoder now takes in the current PC by reference to the getExtMachInst function. It can modify the PC however it needs to (by setting NPC to be the PC + instruction length, for instance). This could be improved since the CPU doesn't know if the PC was modified and always has to write it back. ISA parser: To support the new API, all PC related operand types were removed from the parser and replaced with a PCState type. There are two warts on this implementation. First, as with all the other operand types, the PCState still has to have a valid operand type even though it doesn't use it. Second, using syntax like PCS.npc(target) doesn't work for two reasons, this looks like the syntax for operand type overriding, and the parser can't figure out if you're reading or writing. Instructions that use the PCS operand (which I've consistently called it) need to first read it into a local variable, manipulate it, and then write it back out. Return address stack: The return address stack needed a little extra help because, in the presence of branch delay slots, it has to merge together elements of the return PC and the call PC. To handle that, a buildRetPC utility function was added. There are basically only two versions in all the ISAs, but it didn't seem short enough to put into the generic ISA directory. Also, the branch predictor code in O3 and InOrder were adjusted so that they always store the PC of the actual call instruction in the RAS, not the next PC. If the call instruction is a microop, the next PC refers to the next microop in the same macroop which is probably not desirable. The buildRetPC function advances the PC intelligently to the next macroop (in an ISA specific way) so that that case works. Change in stats: There were no change in stats except in MIPS and SPARC in the O3 model. MIPS runs in about 9% fewer ticks. SPARC runs with 30%-50% fewer ticks, which could likely be improved further by setting call/return instruction flags and taking advantage of the RAS. TODO: Add != operators to the PCState classes, defined trivially to be !(a==b). Smooth out places where PCs are split apart, passed around, and put back together later. I think this might happen in SPARC's fault code. Add ISA specific constructors that allow setting PC elements without calling a bunch of accessors. Try to eliminate the need for the branching() function. Factor out Alpha's PAL mode pc bit into a separate flag field, and eliminate places where it's blindly masked out or tested in the PC. diff 7707:e5b6f1157be3 Sat Oct 16 02:57:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> GetArgument: Rework getArgument so that X86_FS compiles again. When no size is specified for an argument, push the decision about what size to use into the ISA by passing a size of -1. diff 7693:f1db1000d957 Fri Oct 01 17:02:00 EDT 2010 Ali Saidi <Ali.Saidi@ARM.com> Debug: Implement getArgument() and function skipping for ARM. In the process make add skipFuction() to handle isa specific function skipping instead of ifdefs and other ugliness. For almost all ABIs, 64 bit arguments can only start in even registers. Size is now passed to getArgument() so that 32 bit systems can make decisions about register selection for 64 bit arguments. The number argument is now passed by reference because getArgument() will need to change it based on the size of the argument and the current argument number. For ARM, if the argument number is odd and a 64-bit register is requested the number must first be incremented to because all 64 bit arguments are passed in an even argument register. Then the number will be incremented again to access both halves of the argument. diff 7650:42684e4656e6 Wed Aug 25 20:10:00 EDT 2010 Ali Saidi <Ali.Saidi@ARM.com> ARM: Limited implementation of dprintk. Does not work with vfp arguments or arguments passed on the stack. diff 7111:ee902ae075bb Wed Jun 02 01:58:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> ARM: Replace the "never" condition with the "unconditional" condition. |
/gem5/src/arch/power/isa/ | ||
H A D | operands.isa | diff 7791:762276cd3cc7 Wed Dec 08 13:33:00 EST 2010 Gabe Black <gblack@eecs.umich.edu> POWER: Take advantage of new PCState syntax. diff 7720:65d338a8dba4 Sun Oct 31 03:07:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> ISA,CPU,etc: Create an ISA defined PC type that abstracts out ISA behaviors. This change is a low level and pervasive reorganization of how PCs are managed in M5. Back when Alpha was the only ISA, there were only 2 PCs to worry about, the PC and the NPC, and the lsb of the PC signaled whether or not you were in PAL mode. As other ISAs were added, we had to add an NNPC, micro PC and next micropc, x86 and ARM introduced variable length instruction sets, and ARM started to keep track of mode bits in the PC. Each CPU model handled PCs in its own custom way that needed to be updated individually to handle the new dimensions of variability, or, in the case of ARMs mode-bit-in-the-pc hack, the complexity could be hidden in the ISA at the ISA implementation's expense. Areas like the branch predictor hadn't been updated to handle branch delay slots or micropcs, and it turns out that had introduced a significant (10s of percent) performance bug in SPARC and to a lesser extend MIPS. Rather than perpetuate the problem by reworking O3 again to handle the PC features needed by x86, this change was introduced to rework PC handling in a more modular, transparent, and hopefully efficient way. PC type: Rather than having the superset of all possible elements of PC state declared in each of the CPU models, each ISA defines its own PCState type which has exactly the elements it needs. A cross product of canned PCState classes are defined in the new "generic" ISA directory for ISAs with/without delay slots and microcode. These are either typedef-ed or subclassed by each ISA. To read or write this structure through a *Context, you use the new pcState() accessor which reads or writes depending on whether it has an argument. If you just want the address of the current or next instruction or the current micro PC, you can get those through read-only accessors on either the PCState type or the *Contexts. These are instAddr(), nextInstAddr(), and microPC(). Note the move away from readPC. That name is ambiguous since it's not clear whether or not it should be the actual address to fetch from, or if it should have extra bits in it like the PAL mode bit. Each class is free to define its own functions to get at whatever values it needs however it needs to to be used in ISA specific code. Eventually Alpha's PAL mode bit could be moved out of the PC and into a separate field like ARM. These types can be reset to a particular pc (where npc = pc + sizeof(MachInst), nnpc = npc + sizeof(MachInst), upc = 0, nupc = 1 as appropriate), printed, serialized, and compared. There is a branching() function which encapsulates code in the CPU models that checked if an instruction branched or not. Exactly what that means in the context of branch delay slots which can skip an instruction when not taken is ambiguous, and ideally this function and its uses can be eliminated. PCStates also generally know how to advance themselves in various ways depending on if they point at an instruction, a microop, or the last microop of a macroop. More on that later. Ideally, accessing all the PCs at once when setting them will improve performance of M5 even though more data needs to be moved around. This is because often all the PCs need to be manipulated together, and by getting them all at once you avoid multiple function calls. Also, the PCs of a particular thread will have spatial locality in the cache. Previously they were grouped by element in arrays which spread out accesses. Advancing the PC: The PCs were previously managed entirely by the CPU which had to know about PC semantics, try to figure out which dimension to increment the PC in, what to set NPC/NNPC, etc. These decisions are best left to the ISA in conjunction with the PC type itself. Because most of the information about how to increment the PC (mainly what type of instruction it refers to) is contained in the instruction object, a new advancePC virtual function was added to the StaticInst class. Subclasses provide an implementation that moves around the right element of the PC with a minimal amount of decision making. In ISAs like Alpha, the instructions always simply assign NPC to PC without having to worry about micropcs, nnpcs, etc. The added cost of a virtual function call should be outweighed by not having to figure out as much about what to do with the PCs and mucking around with the extra elements. One drawback of making the StaticInsts advance the PC is that you have to actually have one to advance the PC. This would, superficially, seem to require decoding an instruction before fetch could advance. This is, as far as I can tell, realistic. fetch would advance through memory addresses, not PCs, perhaps predicting new memory addresses using existing ones. More sophisticated decisions about control flow would be made later on, after the instruction was decoded, and handed back to fetch. If branching needs to happen, some amount of decoding needs to happen to see that it's a branch, what the target is, etc. This could get a little more complicated if that gets done by the predecoder, but I'm choosing to ignore that for now. Variable length instructions: To handle variable length instructions in x86 and ARM, the predecoder now takes in the current PC by reference to the getExtMachInst function. It can modify the PC however it needs to (by setting NPC to be the PC + instruction length, for instance). This could be improved since the CPU doesn't know if the PC was modified and always has to write it back. ISA parser: To support the new API, all PC related operand types were removed from the parser and replaced with a PCState type. There are two warts on this implementation. First, as with all the other operand types, the PCState still has to have a valid operand type even though it doesn't use it. Second, using syntax like PCS.npc(target) doesn't work for two reasons, this looks like the syntax for operand type overriding, and the parser can't figure out if you're reading or writing. Instructions that use the PCS operand (which I've consistently called it) need to first read it into a local variable, manipulate it, and then write it back out. Return address stack: The return address stack needed a little extra help because, in the presence of branch delay slots, it has to merge together elements of the return PC and the call PC. To handle that, a buildRetPC utility function was added. There are basically only two versions in all the ISAs, but it didn't seem short enough to put into the generic ISA directory. Also, the branch predictor code in O3 and InOrder were adjusted so that they always store the PC of the actual call instruction in the RAS, not the next PC. If the call instruction is a microop, the next PC refers to the next microop in the same macroop which is probably not desirable. The buildRetPC function advances the PC intelligently to the next macroop (in an ISA specific way) so that that case works. Change in stats: There were no change in stats except in MIPS and SPARC in the O3 model. MIPS runs in about 9% fewer ticks. SPARC runs with 30%-50% fewer ticks, which could likely be improved further by setting call/return instruction flags and taking advantage of the RAS. TODO: Add != operators to the PCState classes, defined trivially to be !(a==b). Smooth out places where PCs are split apart, passed around, and put back together later. I think this might happen in SPARC's fault code. Add ISA specific constructors that allow setting PC elements without calling a bunch of accessors. Try to eliminate the need for the branching() function. Factor out Alpha's PAL mode pc bit into a separate flag field, and eliminate places where it's blindly masked out or tested in the PC. |
/gem5/src/arch/sparc/ | ||
H A D | isa.cc | diff 7741:340b6f01d69b Thu Nov 11 05:03:00 EST 2010 Gabe Black <gblack@eecs.umich.edu> SPARC: Clean up some historical style issues. diff 7703:90299d921559 Sun Oct 10 23:38:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> SPARC: Make SPARC's ISA's clear function initialize everything it should. Also make it not set some pointers to NULL potentially introducing a memory leak. That should be done in the constructor. |
H A D | registers.hh | diff 7741:340b6f01d69b Thu Nov 11 05:03:00 EST 2010 Gabe Black <gblack@eecs.umich.edu> SPARC: Clean up some historical style issues. diff 7649:a6a6177a5ffa Wed Aug 25 20:10:00 EDT 2010 Min Kyu Jeong <minkyu.jeong@arm.com> ARM: Fixed register flattening logic (FP_Base_DepTag was set too low) When decoding a srs instruction, invalid mode encoding returns invalid instruction. This can happen when garbage instructions are fetched from mispredicted path |
/gem5/src/sim/ | ||
H A D | eventq.hh | diff 7066:5389824adf3d Sun Apr 18 16:23:00 EDT 2010 Nathan Binkert <nate@binkert.org> event: Allow EventWrapper to take an object reference diff 7063:c0ea4df1ddab Thu Apr 15 19:24:00 EDT 2010 Nathan Binkert <nate@binkert.org> eventq: move EventQueue constructor to cc file Also make copy constructor and assignment operator private. diff 7060:ad784e759a74 Fri Apr 02 18:28:00 EDT 2010 Nathan Binkert <nate@binkert.org> eventq: allow an implicit cast from an EventManager to an EventQueue * diff 7059:4fb4eeb5f412 Fri Apr 02 18:28:00 EDT 2010 Nathan Binkert <nate@binkert.org> eventq: Clean up some flags - Make the initialized flag always available, not just in debug mode. - Make the Initialized flag actually use several bits so it is very unlikely that something that's uninitialized accidentally looks initialized. - Add an initialized() function that tells you if the current event is indeed initialized. - Clear the flags on delete so it can't be accidentally thought of as initialized. - Fix getFlags assert statement. "How did this ever work?" diff 7058:5c7416199efd Fri Apr 02 18:28:00 EDT 2010 Nathan Binkert <nate@binkert.org> eventq: Make priorities just an integer instead of an enum. Symbolic names should still be used, but this makes it easier to do things like: Event::Priority MyObject_Pri = Event::Default_Pri + 1 Remember that higher numbers are lower priority (should we fix this?) diff 7005:3d5c4acb6015 Fri Mar 12 20:31:00 EST 2010 Nathan Binkert <nate@binkert.org> eventq: rearrange a little bit so I can add some stuff diff 7004:b9e4f8a3fea7 Fri Mar 12 20:31:00 EST 2010 Nathan Binkert <nate@binkert.org> eventq: remove some unused includes diff 6982:16963bc29912 Fri Feb 26 21:09:00 EST 2010 Nathan Binkert <nate@binkert.org> events: Give EventWrapped a default name and description |
H A D | init.hh | diff 7674:8e3734851770 Thu Sep 09 17:15:00 EDT 2010 Nathan Binkert <nate@binkert.org> init: don't build files that centralize python and swig code Instead of putting all object files into m5/object/__init__.py, interrogate the importer to find out what should be imported. Instead of creating a single file that lists all of the embedded python modules, use static object construction to put those objects onto a list. Do something similar for embedded swig (C++) code. diff 7502:3ef7ff12c788 Wed Jul 21 18:53:00 EDT 2010 Nathan Binkert <nate@binkert.org> python: Add mechanism to override code compiled into the exectuable If the user sets the environment variable M5_OVERRIDE_PY_SOURCE to True, then imports that would normally find python code compiled into the executable will instead first check in the absolute location where the code was found during the build of the executable. This only works for files in the src (or extras) directories, not automatically generated files. This is a developer feature! |
H A D | serialize.cc | diff 7576:4154f3e1edae Mon Aug 23 12:18:00 EDT 2010 Ali Saidi <Ali.Saidi@ARM.com> Compiler: Fixes for GCC 4.5. diff 7532:3f6413fc37a2 Tue Aug 17 08:17:00 EDT 2010 Steve Reinhardt <steve.reinhardt@amd.com> sim: revamp unserialization procedure Replace direct call to unserialize() on each SimObject with a pair of calls for better control over initialization in both ckpt and non-ckpt cases. If restoring from a checkpoint, loadState(ckpt) is called on each SimObject. The default implementation simply calls unserialize() if there is a corresponding checkpoint section, so we get backward compatibility for existing objects. However, objects can override loadState() to get other behaviors, e.g., doing other programmed initializations after unserialize(), or complaining if no checkpoint section is found. (Note that the default warning for a missing checkpoint section is now gone.) If not restoring from a checkpoint, we call the new initState() method on each SimObject instead. This provides a hook for state initializations that are only required when *not* restoring from a checkpoint. Given this new framework, do some cleanup of LiveProcess subclasses and X86System, which were (in some cases) emulating initState() behavior in startup via a local flag or (in other cases) erroneously doing initializations in startup() that clobbered state loaded earlier by unserialize(). diff 7494:249f174e6f37 Tue Jul 06 01:57:00 EDT 2010 Steve Reinhardt <steve.reinhardt@amd.com> checkpointing: another small overload fix On Nate's advice, overload 'char' as well as 'signed char' and 'unsigned char'. diff 7491:e8ff1fb26dbc Tue Jul 06 00:39:00 EDT 2010 Steve Reinhardt <steve.reinhardt@amd.com> checkpointing: minor cleanup. Move some static checkpoint stuff into the Checkpoint object namespace. diff 7490:1803e46ddf57 Tue Jul 06 00:39:00 EDT 2010 Steve Reinhardt <steve.reinhardt@amd.com> checkpointing: fix minor bug Somehow we now need to explicitly specialize on 'signed char' and not just 'char' to catch cases like int8_t diff 6820:2980bd04e6df Wed Jan 20 01:03:00 EST 2010 Lisa Hsu <Lisa.Hsu@amd.com> util: do checkpoint aggregation more cleanly, fix last changeset. 1) Move alpha-specific code out of page_table.cc:serialize(). 2) Begin serializing M5_pid and unserializing it, but adding an function to do optional paramIn so that old checkpoints don't need to be fixed up. 3) Fix up alpha startup code so that the unserialized M5_pid value is properly written to DTB_IPR_ASN. 4) Fix the memory unserialize that I forgot somehow in the last changeset. 5) Add in an agg_se.py to handle aggregated checkpoints. --bench foo-bar plus positional arguments foo bar are the only changes in usage from se.py. Note this aggregation stuff has only been tested for Alpha and nothing else, though it should take a very minimal amount of work to get it to work with another ISA. diff 6818:5a0e3a283826 Mon Jan 18 17:30:00 EST 2010 Lisa Hsu <Lisa.Hsu@amd.com> util: make a generic checkpoint aggregator that can aggregate different cpts into one multi-programmed cpt. Make minor changes to serialization/unserialization to get it to work properly. Note that checkpoints were made with a comment at the beginning with // - this must be changed to ## to work properly with the python config parser in the aggregator. |
/gem5/src/mem/cache/tags/ | ||
H A D | SConscript | diff 7768:cdb18c1b51ea Fri Nov 19 19:00:00 EST 2010 Ali Saidi <Ali.Saidi@ARM.com> SCons: Support building without an ISA diff 6976:1d7008e14da6 Tue Feb 23 12:33:00 EST 2010 Lisa Hsu <Lisa.Hsu@amd.com> cache: pull CacheSet out of LRU so that other tags can use associative sets. |
/gem5/src/mem/ruby/slicc_interface/ | ||
H A D | Controller.py | diff 7019:a49fd5febdce Mon Mar 22 00:22:00 EDT 2010 Brad Beckmann <Brad.Beckmann@amd.com> ruby: Added copyright to many Ruby *.py files 6876:a658c315512c Fri Jan 29 23:29:00 EST 2010 Steve Reinhardt <steve.reinhardt@amd.com> ruby: Convert most Ruby objects to M5 SimObjects. The necessary companion conversion of Ruby objects generated by SLICC are converted to M5 SimObjects in the following patch, so this patch alone does not compile. Conversion of Garnet network models is also handled in a separate patch; that code is temporarily disabled from compiling to allow testing of interim code. |
/gem5/src/python/m5/ | ||
H A D | simulate.py | diff 7534:c76a14014c27 Tue Aug 17 08:49:00 EDT 2010 Steve Reinhardt <steve.reinhardt@amd.com> misc: add some AMD copyright notices Meant to add these with the previous batch of csets. diff 7532:3f6413fc37a2 Tue Aug 17 08:17:00 EDT 2010 Steve Reinhardt <steve.reinhardt@amd.com> sim: revamp unserialization procedure Replace direct call to unserialize() on each SimObject with a pair of calls for better control over initialization in both ckpt and non-ckpt cases. If restoring from a checkpoint, loadState(ckpt) is called on each SimObject. The default implementation simply calls unserialize() if there is a corresponding checkpoint section, so we get backward compatibility for existing objects. However, objects can override loadState() to get other behaviors, e.g., doing other programmed initializations after unserialize(), or complaining if no checkpoint section is found. (Note that the default warning for a missing checkpoint section is now gone.) If not restoring from a checkpoint, we call the new initState() method on each SimObject instead. This provides a hook for state initializations that are only required when *not* restoring from a checkpoint. Given this new framework, do some cleanup of LiveProcess subclasses and X86System, which were (in some cases) emulating initState() behavior in startup via a local flag or (in other cases) erroneously doing initializations in startup() that clobbered state loaded earlier by unserialize(). diff 7531:f5e86115a07a Tue Aug 17 08:17:00 EDT 2010 Steve Reinhardt <steve.reinhardt@amd.com> sim: fold checkpoint restore code into instantiate() The separate restoreCheckpoint() call is gone; just pass the checkpoint dir as an optional arg to instantiate(). This change is a precursor to some more extensive reworking of the startup code. diff 7528:6efc3672733b Tue Aug 17 08:11:00 EDT 2010 Steve Reinhardt <steve.reinhardt@amd.com> sim: clean up child handling The old code for handling SimObject children was kind of messy, with children stored both in _values and _children, and inconsistent and potentially buggy handling of SimObject vectors. Now children are always stored in _children, and SimObject vectors are consistently handled using the SimObjectVector class. Also, by deferring the parenting of SimObject-valued parameters until the end (instead of doing it at assignment), we eliminate the hole where one could assign a vector of SimObjects to a parameter then append to that vector, with the appended objects never getting parented properly. This patch induces small stats changes in tests with data races due to changes in the object creation & initialization order. The new code does object vectors in order and so should be more stable. diff 7527:fe90827a663f Tue Aug 17 08:08:00 EDT 2010 Steve Reinhardt <steve.reinhardt@amd.com> sim: move iterating over SimObjects into Python. diff 7525:722f2ad014a7 Tue Aug 17 08:06:00 EDT 2010 Steve Reinhardt <steve.reinhardt@amd.com> sim: make Python Root object a singleton Enforce that the Python Root SimObject is instantiated only once. The C++ Root object already panics if more than one is created. This change avoids the need to track what the root object is, since it's available from Root.getInstance() (if it exists). It's now redundant to have the user pass the root object to functions like instantiate(), checkpoint(), and restoreCheckpoint(), so that arg is gone. Users who use configs/common/Simulate.py should not notice. diff 7492:acc1fbbef239 Tue Jul 06 00:39:00 EDT 2010 Steve Reinhardt <steve.reinhardt@amd.com> sim: fold StartupCallback into SimObject There used to be a reason to have StartupCallback be a separate object, but not any more. Now it's just confusing. diff 6977:039202aafc0d Tue Feb 23 12:33:00 EST 2010 Lisa Hsu <Lisa.Hsu@amd.com> stats: this makes some fixes to AverageStat and AverageVector. Also, make Formulas work on AverageVector. First, Stat::Average (and thus Stats::AverageVector) was broken when coming out of a checkpoint and on resets, this fixes that. Formulas also didn't work with AverageVector, but added support for that. |
/gem5/src/arch/ | ||
H A D | SConscript | diff 7799:5d0f62927d75 Mon Dec 20 16:24:00 EST 2010 Gabe Black <gblack@eecs.umich.edu> Style: Replace some tabs with spaces. diff 7756:846fb3ffe0dc Mon Nov 15 15:04:00 EST 2010 Ali Saidi <Ali.Saidi@ARM.com> SCons: Cleanup SCons output during compile diff 7573:ef798deb9a02 Sun Aug 22 21:42:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> X86: Get rid of unused file arguments.hh. diff 6998:188393fbc381 Wed Mar 10 18:39:00 EST 2010 Nathan Binkert <nate@binkert.org> scons: import ply to work around scons sys.path weirdness diff 6994:c6951099a1cb Fri Feb 26 21:14:00 EST 2010 Nathan Binkert <nate@binkert.org> cpu_models: get rid of cpu_models.py and move the stuff into SCons diff 6993:a24d88e850e2 Fri Feb 26 21:14:00 EST 2010 Nathan Binkert <nate@binkert.org> isa_parser: Make SCons import the isa_parser this is instead of forking a new interpreter |
H A D | isa_parser.py | diff 7788:4dd870e2c91d Wed Dec 08 02:08:00 EST 2010 Gabe Black <gblack@eecs.umich.edu> ISA: Get the parser to support pc state components more elegantly. diff 7756:846fb3ffe0dc Mon Nov 15 15:04:00 EST 2010 Ali Saidi <Ali.Saidi@ARM.com> SCons: Cleanup SCons output during compile diff 7720:65d338a8dba4 Sun Oct 31 03:07:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> ISA,CPU,etc: Create an ISA defined PC type that abstracts out ISA behaviors. This change is a low level and pervasive reorganization of how PCs are managed in M5. Back when Alpha was the only ISA, there were only 2 PCs to worry about, the PC and the NPC, and the lsb of the PC signaled whether or not you were in PAL mode. As other ISAs were added, we had to add an NNPC, micro PC and next micropc, x86 and ARM introduced variable length instruction sets, and ARM started to keep track of mode bits in the PC. Each CPU model handled PCs in its own custom way that needed to be updated individually to handle the new dimensions of variability, or, in the case of ARMs mode-bit-in-the-pc hack, the complexity could be hidden in the ISA at the ISA implementation's expense. Areas like the branch predictor hadn't been updated to handle branch delay slots or micropcs, and it turns out that had introduced a significant (10s of percent) performance bug in SPARC and to a lesser extend MIPS. Rather than perpetuate the problem by reworking O3 again to handle the PC features needed by x86, this change was introduced to rework PC handling in a more modular, transparent, and hopefully efficient way. PC type: Rather than having the superset of all possible elements of PC state declared in each of the CPU models, each ISA defines its own PCState type which has exactly the elements it needs. A cross product of canned PCState classes are defined in the new "generic" ISA directory for ISAs with/without delay slots and microcode. These are either typedef-ed or subclassed by each ISA. To read or write this structure through a *Context, you use the new pcState() accessor which reads or writes depending on whether it has an argument. If you just want the address of the current or next instruction or the current micro PC, you can get those through read-only accessors on either the PCState type or the *Contexts. These are instAddr(), nextInstAddr(), and microPC(). Note the move away from readPC. That name is ambiguous since it's not clear whether or not it should be the actual address to fetch from, or if it should have extra bits in it like the PAL mode bit. Each class is free to define its own functions to get at whatever values it needs however it needs to to be used in ISA specific code. Eventually Alpha's PAL mode bit could be moved out of the PC and into a separate field like ARM. These types can be reset to a particular pc (where npc = pc + sizeof(MachInst), nnpc = npc + sizeof(MachInst), upc = 0, nupc = 1 as appropriate), printed, serialized, and compared. There is a branching() function which encapsulates code in the CPU models that checked if an instruction branched or not. Exactly what that means in the context of branch delay slots which can skip an instruction when not taken is ambiguous, and ideally this function and its uses can be eliminated. PCStates also generally know how to advance themselves in various ways depending on if they point at an instruction, a microop, or the last microop of a macroop. More on that later. Ideally, accessing all the PCs at once when setting them will improve performance of M5 even though more data needs to be moved around. This is because often all the PCs need to be manipulated together, and by getting them all at once you avoid multiple function calls. Also, the PCs of a particular thread will have spatial locality in the cache. Previously they were grouped by element in arrays which spread out accesses. Advancing the PC: The PCs were previously managed entirely by the CPU which had to know about PC semantics, try to figure out which dimension to increment the PC in, what to set NPC/NNPC, etc. These decisions are best left to the ISA in conjunction with the PC type itself. Because most of the information about how to increment the PC (mainly what type of instruction it refers to) is contained in the instruction object, a new advancePC virtual function was added to the StaticInst class. Subclasses provide an implementation that moves around the right element of the PC with a minimal amount of decision making. In ISAs like Alpha, the instructions always simply assign NPC to PC without having to worry about micropcs, nnpcs, etc. The added cost of a virtual function call should be outweighed by not having to figure out as much about what to do with the PCs and mucking around with the extra elements. One drawback of making the StaticInsts advance the PC is that you have to actually have one to advance the PC. This would, superficially, seem to require decoding an instruction before fetch could advance. This is, as far as I can tell, realistic. fetch would advance through memory addresses, not PCs, perhaps predicting new memory addresses using existing ones. More sophisticated decisions about control flow would be made later on, after the instruction was decoded, and handed back to fetch. If branching needs to happen, some amount of decoding needs to happen to see that it's a branch, what the target is, etc. This could get a little more complicated if that gets done by the predecoder, but I'm choosing to ignore that for now. Variable length instructions: To handle variable length instructions in x86 and ARM, the predecoder now takes in the current PC by reference to the getExtMachInst function. It can modify the PC however it needs to (by setting NPC to be the PC + instruction length, for instance). This could be improved since the CPU doesn't know if the PC was modified and always has to write it back. ISA parser: To support the new API, all PC related operand types were removed from the parser and replaced with a PCState type. There are two warts on this implementation. First, as with all the other operand types, the PCState still has to have a valid operand type even though it doesn't use it. Second, using syntax like PCS.npc(target) doesn't work for two reasons, this looks like the syntax for operand type overriding, and the parser can't figure out if you're reading or writing. Instructions that use the PCS operand (which I've consistently called it) need to first read it into a local variable, manipulate it, and then write it back out. Return address stack: The return address stack needed a little extra help because, in the presence of branch delay slots, it has to merge together elements of the return PC and the call PC. To handle that, a buildRetPC utility function was added. There are basically only two versions in all the ISAs, but it didn't seem short enough to put into the generic ISA directory. Also, the branch predictor code in O3 and InOrder were adjusted so that they always store the PC of the actual call instruction in the RAS, not the next PC. If the call instruction is a microop, the next PC refers to the next microop in the same macroop which is probably not desirable. The buildRetPC function advances the PC intelligently to the next macroop (in an ISA specific way) so that that case works. Change in stats: There were no change in stats except in MIPS and SPARC in the O3 model. MIPS runs in about 9% fewer ticks. SPARC runs with 30%-50% fewer ticks, which could likely be improved further by setting call/return instruction flags and taking advantage of the RAS. TODO: Add != operators to the PCState classes, defined trivially to be !(a==b). Smooth out places where PCs are split apart, passed around, and put back together later. I think this might happen in SPARC's fault code. Add ISA specific constructors that allow setting PC elements without calling a bunch of accessors. Try to eliminate the need for the branching() function. Factor out Alpha's PAL mode pc bit into a separate flag field, and eliminate places where it's blindly masked out or tested in the PC. diff 7092:fbdf4fca0844 Wed Jun 02 01:57:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> ARM: Fix custom writer/reader code for non indexed operands. diff 6992:0288d8e8b192 Fri Feb 26 21:14:00 EST 2010 Nathan Binkert <nate@binkert.org> isa_parser: move the operand map stuff into the ISAParser class. diff 6991:be03f7cea1f9 Fri Feb 26 21:14:00 EST 2010 Nathan Binkert <nate@binkert.org> isa_parser: move more support functions into the ISAParser class diff 6990:83759f72c7de Fri Feb 26 21:14:00 EST 2010 Nathan Binkert <nate@binkert.org> isa_parser: move more stuff into the ISAParser class diff 6989:73afd9458692 Fri Feb 26 21:14:00 EST 2010 Nathan Binkert <nate@binkert.org> isa_parser: move the formatMap and exportContext into the ISAParser class diff 6988:55b7e9ba3b4f Fri Feb 26 21:14:00 EST 2010 Nathan Binkert <nate@binkert.org> isa_parser: Make stack objects class members instead of globals diff 6987:31ba8b062d08 Fri Feb 26 21:14:00 EST 2010 Nathan Binkert <nate@binkert.org> isa_parser: add a debug variable that changes how errors are reported. This allows us to get tracebacks in certain cases where they're more useful than our error message. |
/gem5/src/mem/ruby/system/ | ||
H A D | RubyPort.hh | diff 7055:4e24742201d7 Fri Apr 02 14:20:00 EDT 2010 Nathan Binkert <nate@binkert.org> ruby: get "using namespace" out of headers In addition to obvious changes, this required a slight change to the slicc grammar to allow types with :: in them. Otherwise slicc barfs on std::string which we need for the headers that slicc generates. diff 7039:bc0b6ea676b5 Mon Mar 22 21:43:00 EDT 2010 Nathan Binkert <nate@binkert.org> ruby: style pass diff 6922:1620cffaa3b6 Fri Jan 29 23:29:00 EST 2010 Brad Beckmann <Brad.Beckmann@amd.com> ruby: Removed static members in RubyPort including hitcallback Removed static members in RubyPort and removed the ruby request unique id. diff 6899:f8057af86bf7 Fri Jan 29 23:29:00 EST 2010 Brad Beckmann <Brad.Beckmann@amd.com> ruby: added the GEMS ruby tester diff 6893:9cdf9b65d946 Fri Jan 29 23:29:00 EST 2010 Brad Beckmann <Brad.Beckmann@amd.com> ruby: FS support using the new configuration system diff 6882:898047a3672c Fri Jan 29 23:29:00 EST 2010 Brad Beckmann <Brad.Beckmann@amd.com> ruby: Ruby changes required to use the python config system This patch includes the necessary changes to connect ruby objects using the python configuration system. Mainly it consists of removing unnecessary ruby object pointers and connecting the necessary object pointers using the generated param objects. This patch includes the slicc changes necessary to connect generated ruby objects together using the python configuraiton system. diff 6876:a658c315512c Fri Jan 29 23:29:00 EST 2010 Steve Reinhardt <steve.reinhardt@amd.com> ruby: Convert most Ruby objects to M5 SimObjects. The necessary companion conversion of Ruby objects generated by SLICC are converted to M5 SimObjects in the following patch, so this patch alone does not compile. Conversion of Garnet network models is also handled in a separate patch; that code is temporarily disabled from compiling to allow testing of interim code. |
/gem5/src/arch/x86/isa/microops/ | ||
H A D | regop.isa | diff 7789:f455790bcd47 Wed Dec 08 03:27:00 EST 2010 Gabe Black <gblack@eecs.umich.edu> X86: Take advantage of new PCState syntax. diff 7720:65d338a8dba4 Sun Oct 31 03:07:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> ISA,CPU,etc: Create an ISA defined PC type that abstracts out ISA behaviors. This change is a low level and pervasive reorganization of how PCs are managed in M5. Back when Alpha was the only ISA, there were only 2 PCs to worry about, the PC and the NPC, and the lsb of the PC signaled whether or not you were in PAL mode. As other ISAs were added, we had to add an NNPC, micro PC and next micropc, x86 and ARM introduced variable length instruction sets, and ARM started to keep track of mode bits in the PC. Each CPU model handled PCs in its own custom way that needed to be updated individually to handle the new dimensions of variability, or, in the case of ARMs mode-bit-in-the-pc hack, the complexity could be hidden in the ISA at the ISA implementation's expense. Areas like the branch predictor hadn't been updated to handle branch delay slots or micropcs, and it turns out that had introduced a significant (10s of percent) performance bug in SPARC and to a lesser extend MIPS. Rather than perpetuate the problem by reworking O3 again to handle the PC features needed by x86, this change was introduced to rework PC handling in a more modular, transparent, and hopefully efficient way. PC type: Rather than having the superset of all possible elements of PC state declared in each of the CPU models, each ISA defines its own PCState type which has exactly the elements it needs. A cross product of canned PCState classes are defined in the new "generic" ISA directory for ISAs with/without delay slots and microcode. These are either typedef-ed or subclassed by each ISA. To read or write this structure through a *Context, you use the new pcState() accessor which reads or writes depending on whether it has an argument. If you just want the address of the current or next instruction or the current micro PC, you can get those through read-only accessors on either the PCState type or the *Contexts. These are instAddr(), nextInstAddr(), and microPC(). Note the move away from readPC. That name is ambiguous since it's not clear whether or not it should be the actual address to fetch from, or if it should have extra bits in it like the PAL mode bit. Each class is free to define its own functions to get at whatever values it needs however it needs to to be used in ISA specific code. Eventually Alpha's PAL mode bit could be moved out of the PC and into a separate field like ARM. These types can be reset to a particular pc (where npc = pc + sizeof(MachInst), nnpc = npc + sizeof(MachInst), upc = 0, nupc = 1 as appropriate), printed, serialized, and compared. There is a branching() function which encapsulates code in the CPU models that checked if an instruction branched or not. Exactly what that means in the context of branch delay slots which can skip an instruction when not taken is ambiguous, and ideally this function and its uses can be eliminated. PCStates also generally know how to advance themselves in various ways depending on if they point at an instruction, a microop, or the last microop of a macroop. More on that later. Ideally, accessing all the PCs at once when setting them will improve performance of M5 even though more data needs to be moved around. This is because often all the PCs need to be manipulated together, and by getting them all at once you avoid multiple function calls. Also, the PCs of a particular thread will have spatial locality in the cache. Previously they were grouped by element in arrays which spread out accesses. Advancing the PC: The PCs were previously managed entirely by the CPU which had to know about PC semantics, try to figure out which dimension to increment the PC in, what to set NPC/NNPC, etc. These decisions are best left to the ISA in conjunction with the PC type itself. Because most of the information about how to increment the PC (mainly what type of instruction it refers to) is contained in the instruction object, a new advancePC virtual function was added to the StaticInst class. Subclasses provide an implementation that moves around the right element of the PC with a minimal amount of decision making. In ISAs like Alpha, the instructions always simply assign NPC to PC without having to worry about micropcs, nnpcs, etc. The added cost of a virtual function call should be outweighed by not having to figure out as much about what to do with the PCs and mucking around with the extra elements. One drawback of making the StaticInsts advance the PC is that you have to actually have one to advance the PC. This would, superficially, seem to require decoding an instruction before fetch could advance. This is, as far as I can tell, realistic. fetch would advance through memory addresses, not PCs, perhaps predicting new memory addresses using existing ones. More sophisticated decisions about control flow would be made later on, after the instruction was decoded, and handed back to fetch. If branching needs to happen, some amount of decoding needs to happen to see that it's a branch, what the target is, etc. This could get a little more complicated if that gets done by the predecoder, but I'm choosing to ignore that for now. Variable length instructions: To handle variable length instructions in x86 and ARM, the predecoder now takes in the current PC by reference to the getExtMachInst function. It can modify the PC however it needs to (by setting NPC to be the PC + instruction length, for instance). This could be improved since the CPU doesn't know if the PC was modified and always has to write it back. ISA parser: To support the new API, all PC related operand types were removed from the parser and replaced with a PCState type. There are two warts on this implementation. First, as with all the other operand types, the PCState still has to have a valid operand type even though it doesn't use it. Second, using syntax like PCS.npc(target) doesn't work for two reasons, this looks like the syntax for operand type overriding, and the parser can't figure out if you're reading or writing. Instructions that use the PCS operand (which I've consistently called it) need to first read it into a local variable, manipulate it, and then write it back out. Return address stack: The return address stack needed a little extra help because, in the presence of branch delay slots, it has to merge together elements of the return PC and the call PC. To handle that, a buildRetPC utility function was added. There are basically only two versions in all the ISAs, but it didn't seem short enough to put into the generic ISA directory. Also, the branch predictor code in O3 and InOrder were adjusted so that they always store the PC of the actual call instruction in the RAS, not the next PC. If the call instruction is a microop, the next PC refers to the next microop in the same macroop which is probably not desirable. The buildRetPC function advances the PC intelligently to the next macroop (in an ISA specific way) so that that case works. Change in stats: There were no change in stats except in MIPS and SPARC in the O3 model. MIPS runs in about 9% fewer ticks. SPARC runs with 30%-50% fewer ticks, which could likely be improved further by setting call/return instruction flags and taking advantage of the RAS. TODO: Add != operators to the PCState classes, defined trivially to be !(a==b). Smooth out places where PCs are split apart, passed around, and put back together later. I think this might happen in SPARC's fault code. Add ISA specific constructors that allow setting PC elements without calling a bunch of accessors. Try to eliminate the need for the branching() function. Factor out Alpha's PAL mode pc bit into a separate flag field, and eliminate places where it's blindly masked out or tested in the PC. diff 7719:f299139501f7 Fri Oct 29 05:20:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> X86: Fault on divide by zero instead of panicing. diff 7626:bdd926760470 Mon Aug 23 12:44:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> X86: Get rid of the flagless microop constructor. This will reduce clutter in the source and hopefully speed up compilation. diff 7620:3d8a23caa1ef Mon Aug 23 12:44:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> X86: Consolidate extra microop flags into one parameter. This single parameter replaces the collection of bools that set up various flavors of microops. A flag parameter also allows other flags to be set like the serialize before/after flags, etc., without having to change the constructor. diff 7480:6a854784be4f Fri Jun 25 03:21:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> X86: Fix div2 flag calculation. diff 7087:fb8d5786ff30 Mon May 24 01:44:00 EDT 2010 Nathan Binkert <nate@binkert.org> copyright: Change HP copyright on x86 code to be more friendly diff 7080:c52c581277bf Wed May 12 03:49:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> X86: Actual change that fixes div. How did that happen? diff 7070:abdcb0389716 Sun May 02 03:39:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> X86: Finally fix a division corner case. When doing an unsigned 64 bit division with a divisor that has its most significant bit set, the division code would spill a bit off of the end of a uint64_t trying to shift the dividend into position. This change adds code that handles that case specially by purposefully letting it spill and then going ahead assuming there was a 65th one bit. |
/gem5/src/arch/x86/ | ||
H A D | utility.cc | diff 7720:65d338a8dba4 Sun Oct 31 03:07:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> ISA,CPU,etc: Create an ISA defined PC type that abstracts out ISA behaviors. This change is a low level and pervasive reorganization of how PCs are managed in M5. Back when Alpha was the only ISA, there were only 2 PCs to worry about, the PC and the NPC, and the lsb of the PC signaled whether or not you were in PAL mode. As other ISAs were added, we had to add an NNPC, micro PC and next micropc, x86 and ARM introduced variable length instruction sets, and ARM started to keep track of mode bits in the PC. Each CPU model handled PCs in its own custom way that needed to be updated individually to handle the new dimensions of variability, or, in the case of ARMs mode-bit-in-the-pc hack, the complexity could be hidden in the ISA at the ISA implementation's expense. Areas like the branch predictor hadn't been updated to handle branch delay slots or micropcs, and it turns out that had introduced a significant (10s of percent) performance bug in SPARC and to a lesser extend MIPS. Rather than perpetuate the problem by reworking O3 again to handle the PC features needed by x86, this change was introduced to rework PC handling in a more modular, transparent, and hopefully efficient way. PC type: Rather than having the superset of all possible elements of PC state declared in each of the CPU models, each ISA defines its own PCState type which has exactly the elements it needs. A cross product of canned PCState classes are defined in the new "generic" ISA directory for ISAs with/without delay slots and microcode. These are either typedef-ed or subclassed by each ISA. To read or write this structure through a *Context, you use the new pcState() accessor which reads or writes depending on whether it has an argument. If you just want the address of the current or next instruction or the current micro PC, you can get those through read-only accessors on either the PCState type or the *Contexts. These are instAddr(), nextInstAddr(), and microPC(). Note the move away from readPC. That name is ambiguous since it's not clear whether or not it should be the actual address to fetch from, or if it should have extra bits in it like the PAL mode bit. Each class is free to define its own functions to get at whatever values it needs however it needs to to be used in ISA specific code. Eventually Alpha's PAL mode bit could be moved out of the PC and into a separate field like ARM. These types can be reset to a particular pc (where npc = pc + sizeof(MachInst), nnpc = npc + sizeof(MachInst), upc = 0, nupc = 1 as appropriate), printed, serialized, and compared. There is a branching() function which encapsulates code in the CPU models that checked if an instruction branched or not. Exactly what that means in the context of branch delay slots which can skip an instruction when not taken is ambiguous, and ideally this function and its uses can be eliminated. PCStates also generally know how to advance themselves in various ways depending on if they point at an instruction, a microop, or the last microop of a macroop. More on that later. Ideally, accessing all the PCs at once when setting them will improve performance of M5 even though more data needs to be moved around. This is because often all the PCs need to be manipulated together, and by getting them all at once you avoid multiple function calls. Also, the PCs of a particular thread will have spatial locality in the cache. Previously they were grouped by element in arrays which spread out accesses. Advancing the PC: The PCs were previously managed entirely by the CPU which had to know about PC semantics, try to figure out which dimension to increment the PC in, what to set NPC/NNPC, etc. These decisions are best left to the ISA in conjunction with the PC type itself. Because most of the information about how to increment the PC (mainly what type of instruction it refers to) is contained in the instruction object, a new advancePC virtual function was added to the StaticInst class. Subclasses provide an implementation that moves around the right element of the PC with a minimal amount of decision making. In ISAs like Alpha, the instructions always simply assign NPC to PC without having to worry about micropcs, nnpcs, etc. The added cost of a virtual function call should be outweighed by not having to figure out as much about what to do with the PCs and mucking around with the extra elements. One drawback of making the StaticInsts advance the PC is that you have to actually have one to advance the PC. This would, superficially, seem to require decoding an instruction before fetch could advance. This is, as far as I can tell, realistic. fetch would advance through memory addresses, not PCs, perhaps predicting new memory addresses using existing ones. More sophisticated decisions about control flow would be made later on, after the instruction was decoded, and handed back to fetch. If branching needs to happen, some amount of decoding needs to happen to see that it's a branch, what the target is, etc. This could get a little more complicated if that gets done by the predecoder, but I'm choosing to ignore that for now. Variable length instructions: To handle variable length instructions in x86 and ARM, the predecoder now takes in the current PC by reference to the getExtMachInst function. It can modify the PC however it needs to (by setting NPC to be the PC + instruction length, for instance). This could be improved since the CPU doesn't know if the PC was modified and always has to write it back. ISA parser: To support the new API, all PC related operand types were removed from the parser and replaced with a PCState type. There are two warts on this implementation. First, as with all the other operand types, the PCState still has to have a valid operand type even though it doesn't use it. Second, using syntax like PCS.npc(target) doesn't work for two reasons, this looks like the syntax for operand type overriding, and the parser can't figure out if you're reading or writing. Instructions that use the PCS operand (which I've consistently called it) need to first read it into a local variable, manipulate it, and then write it back out. Return address stack: The return address stack needed a little extra help because, in the presence of branch delay slots, it has to merge together elements of the return PC and the call PC. To handle that, a buildRetPC utility function was added. There are basically only two versions in all the ISAs, but it didn't seem short enough to put into the generic ISA directory. Also, the branch predictor code in O3 and InOrder were adjusted so that they always store the PC of the actual call instruction in the RAS, not the next PC. If the call instruction is a microop, the next PC refers to the next microop in the same macroop which is probably not desirable. The buildRetPC function advances the PC intelligently to the next macroop (in an ISA specific way) so that that case works. Change in stats: There were no change in stats except in MIPS and SPARC in the O3 model. MIPS runs in about 9% fewer ticks. SPARC runs with 30%-50% fewer ticks, which could likely be improved further by setting call/return instruction flags and taking advantage of the RAS. TODO: Add != operators to the PCState classes, defined trivially to be !(a==b). Smooth out places where PCs are split apart, passed around, and put back together later. I think this might happen in SPARC's fault code. Add ISA specific constructors that allow setting PC elements without calling a bunch of accessors. Try to eliminate the need for the branching() function. Factor out Alpha's PAL mode pc bit into a separate flag field, and eliminate places where it's blindly masked out or tested in the PC. diff 7707:e5b6f1157be3 Sat Oct 16 02:57:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> GetArgument: Rework getArgument so that X86_FS compiles again. When no size is specified for an argument, push the decision about what size to use into the ISA by passing a size of -1. diff 7693:f1db1000d957 Fri Oct 01 17:02:00 EDT 2010 Ali Saidi <Ali.Saidi@ARM.com> Debug: Implement getArgument() and function skipping for ARM. In the process make add skipFuction() to handle isa specific function skipping instead of ifdefs and other ugliness. For almost all ABIs, 64 bit arguments can only start in even registers. Size is now passed to getArgument() so that 32 bit systems can make decisions about register selection for 64 bit arguments. The number argument is now passed by reference because getArgument() will need to change it based on the size of the argument and the current argument number. For ARM, if the argument number is odd and a 64-bit register is requested the number must first be incremented to because all 64 bit arguments are passed in an even argument register. Then the number will be incremented again to access both halves of the argument. diff 7629:0f0c231e3e97 Mon Aug 23 19:14:00 EDT 2010 Gabe Black <gblack@eecs.umich.edu> X86: Create a directory for files that define register indexes. This is to help tidy up arch/x86. These files should not be used external to the ISA. diff 7087:fb8d5786ff30 Mon May 24 01:44:00 EDT 2010 Nathan Binkert <nate@binkert.org> copyright: Change HP copyright on x86 code to be more friendly |
Completed in 235 milliseconds