Commit Graph

388 Commits

Author SHA1 Message Date
Stanislav Shwartsman
6c8db0f569 simplify interfaces to DTLB/ITLB 2019-12-09 18:46:36 +00:00
Stanislav Shwartsman
4b66fecaad split Bochs CPU TLB to DTLB and ITLB to avoid aliasing conflicts between them. ~5% speedup measured 2019-12-09 18:37:02 +00:00
Stanislav Shwartsman
96e2c50bef applying SF patch #545 Speling fixes 2019-12-09 16:29:23 +00:00
Stanislav Shwartsman
2ae332cce8 patch by Luigu.B - significantly speedup multi-threaded guest simulation 2019-08-09 19:57:13 +00:00
Stanislav Shwartsman
773f1b7e42 cleanup return value of all instruction handlers 2018-02-16 07:57:32 +00:00
Stanislav Shwartsman
f490b00beb added support for conditional breakpoints in Bochs debugger 2017-08-22 18:47:18 +00:00
Stanislav Shwartsman
9308ad31c6 remove unused param from serveIcacheMiss 2016-02-22 19:57:24 +00:00
Stanislav Shwartsman
9557cafcef revertng commit #12854 because it broke MT simulation with debugger enabled. Until investigted. 2015-12-20 22:44:54 +00:00
Stanislav Shwartsman
a8a325f2f5 #define to enum or inline function convertion 2015-10-09 19:33:36 +00:00
Stanislav Shwartsman
ea3c1c77eb added vmx consistency checks related to recently implemented support for vm-entering shutdown/wait-for-sipi state 2015-10-09 06:18:14 +00:00
Stanislav Shwartsman
cd2129ec3b avoid calling prefetch() each time when linking traces cross page 2015-10-09 05:33:44 +00:00
Stanislav Shwartsman
9f77a6c3b0 full debugger support together with handler-chaining speedups optimization enabled (experimental)
should speedup emulation with debugger enabled
2015-10-09 05:28:47 +00:00
Stanislav Shwartsman
c44cb6ed81 more cases applicable for BX_TLB_ENTRY_OF 2015-09-22 20:10:22 +00:00
Stanislav Shwartsman
be4b73c6d2 extracted tlb specific code to tlb.h; extracted xsave cpuid leaf function to base cpuid class 2015-09-21 13:16:17 +00:00
Stanislav Shwartsman
6252632e31 Fixed segmentation fault that could happen under rare conditions with handlers chaining speedups enabled.
I saw that issue under gcc 4.9.0. for some reason gcc 4.9.0 didn't optimize next handler call in all fpu opcode handlers.
As result, instead of finishing the handler and jumping to next one, the next handler is called blowing up stack.
After some long period stack overflow might occur.

The fix simply limit the max chaining depth to 1000 traces (should be enough)
The same fix should be able to address the stack overflow problem when compiling with -O0 and handlers chaining speedup enabled.
2014-10-15 18:00:04 +00:00
Stanislav Shwartsman
8d1e3b2ac1 Added statistics collection infrastructure in Bochs and
implemented important CPU statistics which were used for Bochs CPU model performance analysis.
old statistics code from paging.cc and cpu.cc is replaced with new infrastructure.

In order to enale statitics collection in Bochs CPU:

- Enable statistics @ compilation time in cpu/cpustats.h
- Dump statistics periodically by adding -dumpstats N into Bochs command line
2014-10-14 15:59:10 +00:00
Stanislav Shwartsman
24cb334304 fixed large code duplication in write_new_stack methods 2014-10-12 18:59:10 +00:00
Stanislav Shwartsman
2fe0aaa472 added configure option for trace linking optimization and mention it in CHANGES 2014-05-01 18:30:23 +00:00
Stanislav Shwartsman
3be7e5884b added lock prefix used info into bx_Instriction_c and use it in disasm 2013-11-08 21:43:21 +00:00
Stanislav Shwartsman
0fd4e3450c update (c) for few files 2013-09-05 18:40:14 +00:00
Stanislav Shwartsman
ef0d2142ab Allow cross-page trace linking again.
The SMC problem was solved in following manner:

 - Every trace linked to another remembers when it was linked (a special timestamp value called traceLinkTimeStamp)
 - When true SMC happens it incremements the traceLinkTimeStamp
 - Jump to the linked trace won't be allowed if traceLinkTimeStamp in the link doesn't match traceLinkTimeStamp

So SMC effectively breaks all trace links and therefore I should not care for them anymore

5%-10% speedup on OS boot benchamarks observed
2013-06-29 10:16:28 +00:00
Stanislav Shwartsman
0276bdfb3e still not allow cross page linking until SMC issue will be solved - cause Win98 crash 2013-06-28 07:51:42 +00:00
Stanislav Shwartsman
c42afb0a2d allow linking of traces cross 4K page boundary 2013-06-23 21:12:03 +00:00
Stanislav Shwartsman
91b3417e57 small bugfix 2013-06-23 15:45:25 +00:00
Stanislav Shwartsman
d38fce8218 preparation for future extension in translate_linear - I would like to return data to caller through tlbEntry 2013-01-27 19:27:30 +00:00
Stanislav Shwartsman
eda28b95f4 unfortunately this change is rquired to make SMAP and SMEP features to work.
I observed ~5% emulation slowdown ... thinking about possible mitigations

this fixes TLB issue with SMAP and SMEP features.
these features introduce a new behavior when page can be inaccessible by System (CPL=0).
Current behavior is accessBits was not supporting it but legacy (from Bochs 2.3.6) was.
The wrong behavior can be observed if user access a user page and system access the same page later.
user access is fine and pass SMEP/SMA checks and stores the translation in TLB.
the system access will hit the TLB and nobody could detect that system cannot access that page.
2013-01-16 17:28:20 +00:00
Stanislav Shwartsman
574b69c81e fixed MSDEV warnings 2012-11-27 15:40:45 +00:00
Stanislav Shwartsman
e397a86ce0 fixed code duplication related to EXT field 2012-09-29 09:31:34 +00:00
Stanislav Shwartsman
40ba9c8d7b introducing new interface for handling CPU events based on vector of events and not on many not related variables. this is very initial implementation which takes into new interface only few events, more will code soon 2012-09-25 09:35:38 +00:00
Stanislav Shwartsman
8044a2bda6 rename i->execute field in the instruction
move victim cache lookup into cache lookup so traces could be linked with victim cache hits directly
2012-09-04 15:45:05 +00:00
Stanislav Shwartsman
e17cffab57 simplify generated code 2012-08-26 15:49:30 +00:00
Stanislav Shwartsman
c41cbe6d56 Link traces over taken branch optimization which makes handlers chaining even more efficient.
I observed 5% speedup in all disk images over 2.6pre1.
The change is safe (passed all regressions) and I will be glad to make it into Bochs 2.6!
2012-08-21 19:58:41 +00:00
Stanislav Shwartsman
ec06475dbf improve x86 hw breakpoint handling 2012-07-11 15:07:54 +00:00
Stanislav Shwartsman
58dde88887 VME is for CPU_LEVEL>=5 only 2012-07-08 18:16:25 +00:00
Stanislav Shwartsman
79628a2f4f fixed VME corner case 2012-06-27 15:09:10 +00:00
Stanislav Shwartsman
bd6330d480 small optimization for debugger 2012-06-03 18:46:20 +00:00
Stanislav Shwartsman
a68ad9a7f6 small code optimization 2012-04-06 09:41:58 +00:00
Stanislav Shwartsman
279c61dc67 updated + fixed instrumentation example for instr histogram, code cleanup in the cpu 2012-03-28 21:11:19 +00:00
Stanislav Shwartsman
25ffaeeea8 fixed VMX issue + small code reorg 2012-03-13 15:18:21 +00:00
Stanislav Shwartsman
864ea23b5b take events handling logic from cpu.cc to new file event.cc 2011-12-28 12:26:45 +00:00
Stanislav Shwartsman
2b8371f2b6 implemented SVM_GIF handling 2011-12-27 20:46:15 +00:00
Stanislav Shwartsman
7f5f917a34 more SVM implementation 2011-12-27 19:42:11 +00:00
Stanislav Shwartsman
e7ed8aca5c move inhibit interrrupts functionality to icount interface 2011-12-21 06:17:45 +00:00
Stanislav Shwartsman
1b9f286945 - New way of CPUs scheduling in SMP mode brings up to 50% speedup to the
SMP emulation. New implementation uses dynamic CPU quantum value and takes
   full advantage of the trace cache. Each emulated processor will execute
   the whole trace before switching to the next processor.
 * It is also safe to use large (up to 16 instructions) quantum values for
   the SMP emulation now and improve performance even further.

The same merge also completely fixes SF bug :
  [3312237] stepN command might be not working properly

Handlers chaining speedups are also supported with SMP emulation now.
2011-09-22 19:38:52 +00:00
Stanislav Shwartsman
f4dbaf1cd8 re-shuffle macros, no impact in general 2011-09-13 17:55:36 +00:00
Stanislav Shwartsman
184837e0ed fixed compilation err with no handlers chaining enabled 2011-09-06 15:41:52 +00:00
Stanislav Shwartsman
96cedbc756 continue handlers-chaining optimization: update time once per trace and not for every instruction 2011-09-06 15:35:39 +00:00
Stanislav Shwartsman
5a350143a5 bug fixes 2011-09-06 13:09:45 +00:00
Stanislav Shwartsman
c67338203c small fixups, code cleanup and reorganization 2011-09-05 17:14:49 +00:00
Stanislav Shwartsman
13feb0772a - 10% emulation speedup with handlers chaining optimization implemented. The
feature is enabled by default when configure with --enable-all-optimizations
    option, to disable handlers chaining speedups configure with
        --disable-handlers-chaining
2011-08-21 14:31:08 +00:00