[Mplayerxp-cvslog] SF.net SVN: mplayerxp:[438] mplayerxp
Brought to you by:
olov
Menu
▾
▴
[Mplayerxp-cvslog] SF.net SVN: mplayerxp:[438] mplayerxp
|
From: <nic...@us...> - 2012-11-23 09:18:37
|
Revision: 438
http://mplayerxp.svn.sourceforge.net/mplayerxp/?rev=438&view=rev
Author: nickols_k
Date: 2012-11-23 09:18:26 +0000 (Fri, 23 Nov 2012)
Log Message:
-----------
Change number of backtraces.
Note: i suspect that my development environment has been virused to add
illegal patch into MPlayerXP.
When i perform transition on to C++ then i found out the strange behaviour
of ld:
- [424] commit didn't break linkage of the project althrough this patch
replaces some short C-names of functions with C++-long names. Therefore,
short C-names like rtsp_session_start() were linked by g++ regardless the fact
that official source-tree of project lost them.
- [437] commit emulates fixing of problem. I had movie which couldn't be played
before this commit. But additional MSG_V() cannot fix anything in player.
Furthermore, player after this commit doesn't print some symbols which it
must print with this patch.
- list of suspect calls with -core.malloc-debug=3 option points me that some
internal contexts of player were substituted by illegal patch. Additional
proof of that is [434] commit. This commit did illegal patch thinner, but not
removed it. Furthemore, each replace of #define with inline function or enum
decreases size of project regardless on long C++-names of functions and data
objects.
Well, one of goal transition project on C++ was using of C++ long-names which
are dynamically changed together with changing number or types of arguments.
But, it seems that there is big aliance against of me and my project. And i
suspect some objects from mplayerhq team who revenge me for fork using dirty
technology. Considering the fact that they hate C++ i'm suspecting that they
coerce somebody to attack my project and modify illegal patch parallel with
my improvements.
Additional observations:
- My local copy of g++ still prints: "warning: 'packed' attribute ignored"
that makes win32loader non working at least. That warning became appears
right after adding so-called small secury holes into structures of project.
- My local copy of /lib32/libgcc_s.so was damaged because demuxer returns -nan
in pts field for each packet. Moreover, upgrade of gcc didn't help to solve
this problem. Thus, i have no working 32-bit development environment.
- My 32-bit development environment (aka: gcc-4.5.4) printed:
--- 8< ----
In file included from ../osdep/cpudetect.h:4:0,
from win32.c:942:
../mplayerxp.h:130:20: error: invalid storage class for function 'escape_player'
../mplayerxp.h:130:20: warning: no previous prototype for 'escape_player'
../mplayerxp.h:135:23: error: invalid storage class for function 'check_pin'
../mplayerxp.h:135:23: warning: no previous prototype for 'check_pin'
--- 8< ---
mplayerxp.h:130:
static inline void escape_player(const char* why,unsigned num_calls) {
mplayerxp.h:135:
static inline MPXP_Rc check_pin(const char* module,unsigned pin1,unsigned pin2) {
--- 8< ---
- Last few days, 32-bit binary of mplayerxp is able to print help only and
ignores command-line arguments.
But who is that (***insert any colorful phrases here***) analog of virus which
attacks my development environment and who is his host(s)?
Thus, i need find out the way to restore the quality of my development enviroment!
Modified Paths:
--------------
mplayerxp/configure
mplayerxp/libao2/ao_sdl.cpp
mplayerxp/osdep/mp_malloc.cpp
Modified: mplayerxp/configure
===================================================================
--- mplayerxp/configure 2012-11-22 18:37:26 UTC (rev 437)
+++ mplayerxp/configure 2012-11-23 09:18:26 UTC (rev 438)
@@ -795,6 +795,7 @@
bsd && add_ldflags "-rdynamic"
enabled dbg23 def_debug='#define MP_DEBUG 1' || def_debug='#undef MP_DEBUG'
+enabled dbg23 && check_cflags "-fbounds-check"
# Checking for VIDIX
enabled vidix && xxrequire2 vidix "vidix/vidix.h vidix/vidixlibxx.h" Vidix -lvidixxx
Modified: mplayerxp/libao2/ao_sdl.cpp
===================================================================
--- mplayerxp/libao2/ao_sdl.cpp 2012-11-22 18:37:26 UTC (rev 437)
+++ mplayerxp/libao2/ao_sdl.cpp 2012-11-23 09:18:26 UTC (rev 438)
@@ -9,6 +9,7 @@
* Thanks to Arpi for nice ringbuffer-code!
*
*/
+#include <algorithm>
#include <stdio.h>
#include <stdlib.h>
@@ -56,8 +57,7 @@
int x;
while(len>0){
if(priv->full_buffers==NUM_BUFS) break;
- x=BUFFSIZE-priv->buf_write_pos;
- if(x>len) x=len;
+ x=std::min(unsigned(len),BUFFSIZE-priv->buf_write_pos);
memcpy(priv->buffer[priv->buf_write]+priv->buf_write_pos,data+len2,x);
len2+=x; len-=x;
priv->buffered_bytes+=x; priv->buf_write_pos+=x;
@@ -77,8 +77,7 @@
int x;
while(len>0){
if(priv->full_buffers==0) break; // no more data buffered!
- x=BUFFSIZE-priv->buf_read_pos;
- if(x>len) x=len;
+ x=std::min(unsigned(len),BUFFSIZE-priv->buf_read_pos);
memcpy(data+len2,priv->buffer[priv->buf_read]+priv->buf_read_pos,x);
SDL_MixAudio(data+len2, data+len2, x, priv->volume);
len2+=x; len-=x;
Modified: mplayerxp/osdep/mp_malloc.cpp
===================================================================
--- mplayerxp/osdep/mp_malloc.cpp 2012-11-22 18:37:26 UTC (rev 437)
+++ mplayerxp/osdep/mp_malloc.cpp 2012-11-23 09:18:26 UTC (rev 438)
@@ -17,11 +17,13 @@
namespace mpxp {
+enum { Max_BackTraces=13 };
+
typedef struct mp_slot_s {
any_t* page_ptr;
size_t size;
size_t ncalls;
- any_t* calls[10];
+ any_t* calls[Max_BackTraces];
}mp_slot_t;
typedef struct mp_slot_container_s {
@@ -218,7 +220,7 @@
if(!slot) {
printf("[__prot_free_append] suspect call found! Can't find slot for address: %p [aligned: %p]\n",ptr,page_ptr);
__prot_print_slots(&priv->mallocs);
- __print_backtrace(10);
+ __print_backtrace(Max_BackTraces);
kill(getpid(), SIGILL);
}
size_t fullsize=app_fullsize(slot->size);
@@ -234,7 +236,7 @@
if(!slot) {
printf("[__prot_realloc_append] suspect call found! Can't find slot for address: %p [aligned: %p]\n",ptr,prot_page_align(ptr));
__prot_print_slots(&priv->mallocs);
- __print_backtrace(10);
+ __print_backtrace(Max_BackTraces);
kill(getpid(), SIGILL);
}
memcpy(rp,ptr,std::min(slot->size,size));
@@ -268,7 +270,7 @@
if(!slot) {
printf("[__prot_free_prepend] suspect call found! Can't find slot for address: %p [aligned: %p]\n",ptr,page_ptr);
__prot_print_slots(&priv->mallocs);
- __print_backtrace(10);
+ __print_backtrace(Max_BackTraces);
kill(getpid(), SIGILL);
}
mprotect(page_ptr,__VM_PAGE_SIZE__,MP_PROT_READ|MP_PROT_WRITE);
@@ -283,7 +285,7 @@
if(!slot) {
printf("[__prot_realloc_prepend] suspect call found! Can't find slot for address: %p [aligned: %p]\n",ptr,pre_page_align(ptr));
__prot_print_slots(&priv->mallocs);
- __print_backtrace(10);
+ __print_backtrace(Max_BackTraces);
kill(getpid(), SIGILL);
}
memcpy(rp,ptr,std::min(slot->size,size));
@@ -324,7 +326,7 @@
rp=malloc(size);
if(rp) {
slot=prot_append_slot(&priv->mallocs,rp,size);
- slot->ncalls=backtrace(slot->calls,10);
+ slot->ncalls=backtrace(slot->calls,Max_BackTraces);
}
return rp;
}
@@ -335,7 +337,7 @@
if(rp) {
mp_slot_t* slot;
slot=prot_append_slot(&priv->mallocs,rp,size);
- slot->ncalls=backtrace(slot->calls,10);
+ slot->ncalls=backtrace(slot->calls,Max_BackTraces);
}
return rp;
}
@@ -351,7 +353,7 @@
MSG_WARN("[bt_realloc] suspect call found! Can't find slot for address: %p\n",ptr);
mp_slot_t* _slot;
_slot=prot_append_slot(&priv->reallocs,ptr,size);
- _slot->ncalls=backtrace(_slot->calls,10);
+ _slot->ncalls=backtrace(_slot->calls,Max_BackTraces);
} else {
slot->page_ptr=rp; // update address after realloc
slot->size=size;
@@ -366,7 +368,7 @@
MSG_WARN("[bt_free] suspect call found! Can't find slot for address: %p\n",ptr);
mp_slot_t* _slot;
_slot=prot_append_slot(&priv->frees,ptr,0);
- _slot->ncalls=backtrace(_slot->calls,10);
+ _slot->ncalls=backtrace(_slot->calls,Max_BackTraces);
return;
}
prot_free_slot(&priv->mallocs,ptr);
@@ -389,7 +391,7 @@
}
}
if(printable) MSG_INFO("%20s",s);
- else for(j=0;j<std::min(c->slots[i].size,size_t(10));j++) {
+ else for(j=0;j<std::min(c->slots[i].size,size_t(Max_BackTraces));j++) {
MSG_INFO("%02X ",(unsigned char)s[j]);
}
MSG_INFO("\n");
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
