|
I am giving example as your requirement.
eg.
CString str = "sample(1,2,3,4,5)",str1,str2;
int curPos= 0;
str1 = str.Tokenize("()",curPos);
Output: str1 = 1,2,3,4,5
|
|
|
|
|
Hey
I used DeleteVolumeMountPoint function to remove a USB drive (E from my machine. Now i want to bring that back. I tried re-booting and that doesn't seems to help.
I think calling SetVolumeMountPoint function will bring it back. But that function needs the GUID of USB drive, which i don't have.
Any ideas how to get the USB drive(volume - E:\) back?
Can i do that via regedit? If so, how?
THanks in advance
|
|
|
|
|
Got a system restore point from before you deleted it?
[If not, why not?  ]
Software rusts. Simon Stephenson, ca 1994.
|
|
|
|
|
If you want to do it programmatically, there is some work to do around the SetVolumeMountPoint function.
The easiest way is to do it from the Control Panel: open the Administrative tools and run Computer management, then select the section Disk management. There you can see all the disks, both removable and not; you should right-click on the USB drive and select Change drive letter and path (or Assign drive letter and path, or similar).
Cheers,
Sauro
|
|
|
|
|
Hey Sauro,
That worked awesome!
Thanks a lot.
|
|
|
|
|
I'm using Visual C++ 6 service pack 6 on XP. According to the documentation for that version, Advanced Windows Extentions (AWE) allows users to access physical memory beyond the 2 GB boundary for a process. There is an example in the documentation:
http://msdn.microsoft.com/en-us/library/aa366531(VS.85).aspx
I've had trouble linking to kernel32.dll to make this work. However, the kernel32.dll(2009) in Windows/system32 has the functions so I used LoadLibrary() to explicitly link to that kernel32.dll. Unfortunately, even on a system with 2.96 GB of RAM, running ONLY the simple example, I cannot get VirtualAlloc() to successfully reserve memory over 1100 MB (it does work at 1100 MB and below). I am using the VirtualAlloc from the 2009 kernel32.dll.
*****
typedef LPVOID (WINAPI *LPVIRT)(LPVOID lpAddress,SIZE_T dwSize,DWORD flAllocationType,DWORD flProtect);
LPVIRT lpVirt;
HMODULE hHandle = LoadLibrary("kernel32.dll");
// Reserve the virtual memory.
lpVirt = (LPVIRT)GetProcAddress(hHandle, "VirtualAlloc");
lpMemReserved = (*lpVirt)( NULL,m_uMBytes*0x100000,MEM_RESERVE | MEM_PHYSICAL,PAGE_READWRITE );
if( lpMemReserved == NULL )
{
TRACE("Cannot reserve memory.\n");
return;
}
*****
With 2.96 GB, shouldn't I be able to reserve more than 1100 MB? How can I do it?
|
|
|
|
|
Hi,
1.
please use PRE tags when showing code snippets, as it improves readability.
2.
on 32-bit Windows, the user part of the virtual address space is limited to 2GB (per process). There is a way to extend that to 3GB, use Google to locate it. I have the impression you didn't do it, since you didn't mention it at all. On 64-bit Windows, the situation obviously is quite different.
3.
within 2GB virtual space, I'm not very surprised you get only some 1.1GB; as your own code and data is in that same 2GB space. I don't know what the default addresses for those would be; a link map should contain all such info. Have a look at the linker switches and ask for a map; then maybe look for more switches if you want to alter the virtual addresses used by code and data.
4.
I haven't used MEM_PHYSICAL yet, and haven't read up on all the doc, however I would be concerned it may also imply you are asking for contiguous physical memory (mind you, I'm not sure here). Finding such amount of contiguous memory is pretty unlikely.
5.
Maybe you should tell a bit about the overall goals; just maybe there is a better approach for what you are after.
|
|
|
|
|
Thank you very much for your prompt reply. My overall goal is to be able to have my application access memory beyond the 2GB boundary so that it can allocate physical memory beyond the 2GB boundary and I am able to allocate 2 to 3 (maybe more?) GBs of memory. One solution you mentioned (the 3GB extension) is described here:
http://msdn.microsoft.com/en-us/library/ms189334(SQL.100).aspx
This solution however affects the allocation of memory for every application on the system, and is not something I really want to do. I just want my particular application to be able to allocate beyond the boundary.
Note that the link above mentions AWE, which I am trying to use. Note that the diagram shows the that memory allocated in physical memory is not contiguous, which addresses your concern. The link posted in my first post is the code I am using to use AWE. As I mentioned, the only difference is that I'm using LoadLibrary to access the allocation and mapping functions. This link discusses AWE in more detail:
http://msdn.microsoft.com/en-us/library/aa366527(VS.85).aspx
Included in description for the AWE functions in the documentation is the following: "Physical pages can reside at any physical address. You should make no assumptions about the contiguity of the physical pages."
So the hitch at this point to making this work is VirtualAlloc(). The previous function in the example, "AllocateUserPhysicalPages()" works fine.
|
|
|
|
|
OK, I haven't thought this through yet, however I have one more suggestion:
put your code in a loop, i.e. try allocating the largest you can get, display the size (and address), do not release it, and repeat this all over and over and over. You might be getting a number of probably decreasing chunks, and the sum of those sizes (and maybe addresses) could tell you something interesting.
|
|
|
|
|
VirtualAlloc never allocate over 2GB in 32bit windows without /3GB option.
In some ordinary OS settings situation, VirtualAlloc usually returns under 1.5GB space for user land memory.
And AWE is for use of physical memory over physical 2GB boundary with /pae boot option.
If you want continuous virtual memory space over 2GB, you must use 64bit windows.
With /pae boot option, msdn document says VirtualAlloc can use over 4GB physical memory, but it would not so ordinary 32bit windows especially xp.
Because microsoft dose not support over 4GB physical memory on 32bit xp.
|
|
|
|
|
You're not allocating memory with that call to VirtualAlloc, you're allocating address space. There's only 2Gb of that per application [1] so the fact that you can't allocate more than ~1Gb in one fell swoop isn't surprising.
To be honest if you really need oodles of contiguous memory (i.e. more than 1Gb) then using a 32 bit OS is probably the wrong way to go - try a 64 bit OS and compiler then you won't have to muck around with low level stuff like this.
Cheers,
Ash
[1] Pedants please note the following exclusions: address space available to user mode code, Win32 only without /3Gb switch in boot.ini
|
|
|
|
|
|
|
You should not have removed you original message.
Your solution is worthless without knowing what the original problem was.
Watched code never compiles.
|
|
|
|
|
Does the C++ standard say anything about which side of the assignment operator is executed first? For example,
int j[1];
int i = 0;
j[i++] = i; Will j[0] be 0 or 1 after this block of code executes? Is this per the C++ standard, or is the result of this code compiler-dependent? In general, what rules govern the order of code execution like this? Is it operator associativity (assignment is right-to-left associative, so j[0] = 0), or operator precedence (post-increment and operator[] both have higher precedence than assignment, so j[0] = 1)? Thanks,
Sounds like somebody's got a case of the Mondays
-Jeff
|
|
|
|
|
You get j[0] = 0; the postfix increment is meant as "use before increment" and it produce the same result than the statements below:
j[i] = i;
i = i + 1;
You can try the following:
int j[1];
int i = 0;
j[i] = j++;
This again will produce j[0] = 0.
The answer to your question is: the increment is applied after the whole statement has been processed, like as you wrote the same statement without the postfix increment (i.e. j[i] = i) and then another statement with only the increment (i.e. i = i + 1).
|
|
|
|
|
I assume you meant to use "i++" on the third line of your second example, but understand what you meant. I really thought that the statements:
int i = 1;
int j = (i) + ++i; would be equivalent to "int j = 3;", but no matter how I apply the parentheses, it always comes out as 4, just as your response predicts it would. Thanks,
Sounds like somebody's got a case of the Mondays
-Jeff
|
|
|
|
|
This would be;
int i = 1;
i = i + 1;
int j = i + i;
|
|
|
|
|
IMO that is irrelevant as you should not even consider writing code like that. The only things you can achieve here is confuse yourself, confuse all future readers, and make debugging tougher than necessary.
If it is
j[i]=i;
i++;
you want, then write exactly that. It makes perfectly clear what is going on, and, at least in an Intel-Windows environment, it does not have any influence on code size or performance.
|
|
|
|
|
Well actually, I am trying to write something like the following where I am copying from a 0-indexed array to a 1-indexed array:
const int src[] = {0, 1};
const size_t srcLen = sizeof(src) / sizeof(int);
int *const dst = new int[srcLen] - 1;
for (size_t i = 0; i < srcLen; ++i)
dst[i + 1] = src[i];NOTE: I have simplified the array copy in the above example... I am actually using an external API function to get the value of the src array elements, which is why I am not just using the statement "const int* const dst = &src[0] - 1;".
The compiler will probably optimize this by keeping the value of i + 1 from the array assignment for the next loop iteration in a register (as opposed to recomputing it), but if I could write the for-loop as:
for (size_t i = 0; i < srcLen; )
dst[++i] = src[i];This would *almost* guarantee that i is incremented only once. Given the initial response, I can't have a single statement with two different values for i, so I'm not able to do what I wanted anyway. It would probably have a higher probability of being optimized if I wrote the loop as:
for (size_t i = 0; i < srcLen; ) {
const int temp = src[i];
dst[++i] = temp;
}Almost every compiler available would put temp in a register instead of allocating it on the stack, but I really don't want to write the loop like that. Guess I'll keep my fingers crossed that the compiler optimizes it! (I don't really care timing-wise as the loop is only iterating over a couple hundred thousand elements one time per run, which is an insignificant amount of time as a fraction of the program's run-time, but why not write efficient code when you can?)
Sounds like somebody's got a case of the Mondays
-Jeff
|
|
|
|
|
OK, now I'll try and answer as good as I can, from memory:
1. in C the statement dst[i++]=i; is undefined, i.e. the exact order of the operation is not specified by the language specification. A compiler might first evaluate the RHS expression, then start working on the LHS; or it could first reduce the LHS to a simple target address, then evaluate the RHS, then assign. A compiler would even be allowed to unroll the loop (say by a factor of 2) and treat both halves differently. I don't think you'd like that.
2. in C++ things are slightly better, and AFAICR the spec got more stringent over time; the compiler builder still has options but is urged to document the choices he made; I have actually seen compiler vendors do it. So for a specific compiler, you might officially get away with such code. You still would not pass my code review though, my style guide would stop you in time. 
Skippums wrote: why not write efficient code when you can?
I agree to some extent. When two styles are possible: if readability is the same, I'll opt for the higher performance variant. If a trade-of is to be made, I'll judge case by case. And if performance is all and everything, I'll put the normal code as a comment and write whatever it takes.
But you have me puzzled even more by your base-0 and base-1 indexing. Can't you just make it all consistent (and I do mean base-0) throughout your app and get rid of the problem entirely?
|
|
|
|
|
I have two arrays being sent to my function from an outside source. The first array is a list of 1-based indices into the the second array that contains the data. When the first array has an index of 0, this means that there is no associated data element in the second array. So, for example:
const size_t indexArray[] = {1, 0, 3, 2, 0};
DATA_TYPE *const dataArray = new DATA_TYPE[3] - 1; Now, as I said, no matter what I do I have to read in the values of dataArray using API calls on a per-element basis. I could very easily copy the values from indexArray into another variable in my C-code, subtracting 1 from each element as I perform the copy. However, I think it is more straightforward (and more efficient) to just get a pointer to the existing indexArray data (using an API call) and subtract one from my dynamically allocated dataArray object as I illustrated in the above code. Using this approach, I can compare each element from indexArray to 0 to test for the special condition, whereas if I did a copy and subtracted one I would have to test against numeric_limits<size_t>::max(), and pray that the max() macro from windows.h wasn't included by some other header file that I am including. I may also need to make the copy loop more complex, if I don't want to assume that the result of (0u - 1u) is the maximum unsigned integer.
Hopefully that explains in better detail why I am not just converting it to 0-based indices.
Sounds like somebody's got a case of the Mondays
-Jeff
|
|
|
|
|
Okay, my mind is completely boggled by this code...
- you allocate an intrinsic array of size whatever
- you decrement the pointer to the array
- you then manually copy another array into this one starting at the same position you allocated the memory at
What's wrong with...
int *src_start = ...
int *src_end = ...
int *p = new int[ src_end - src_start ];
std::copy( src_start, src_end, p );
int *p_to_1_based_alias = p - 1;
If that doesn't at least perform as well as your handrolled loop then trade your compiler in for a new one. You can also get rid of any explicit memory management using a vector.
Another idea would be to just front end the array you get back from the API and convert the index:
template<std::size_t N, typename T>
class N_based_array
{
public:
N_based_array( T *data ) : data_( data ) {}
T &operator[]( std::size_t index )
{
return data_[ index - N ];
}
private:
T *data_;
};
Cheers,
Ash
|
|
|
|
|
I don't get a list of the type I want back from the API function call... I have to call each API method on a per-element basis. Basically, I have a cell-array from Matlab being passed in:
size_t const indexCount = mxGetNumberOfElements(prhs[0]);
size_t const dataSize = mxGetNumberOfElements(prhs[1]);
double const ** data = new double const *[dataSize];
unsigned __int32 const *const indices = static_cast<unsigned __int32 const *>(mxGetData(prhs[0]));
for (size_t i = 0; i < dataSize; ++i) {
data[i] = mxGetPr(mxGetCell(prhs[1], i));
}
--data; That is why I didn't do what you suggested.
Sounds like somebody's got a case of the Mondays
-Jeff
|
|
|
|
|
So what's wrong with the wrapping class solution to convert your array to a 1 based one?
Seeing your code there you could use std::generate on a std::vector to get the same effect as your manual loop. Generally if you're doing low level memory fiddling, pointer arithmetic and looping at the same time in C++ there's usually a simpler way.
Cheers,
Ash
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
