Microsoft Visual C++ Runtime Library error

[FileEX]

Describe the bug Some times mta has been crashed and show error. It's totally random and rare.

To reproduce

Expected behaviour Fix it

Screenshots

Version 1.5.6

Additional context

[Pirulax]

Hm, interesting... My PR had the same problem(#907) if there were a lot of elements in a google::dense_hash_map(in MTA SC referred as CFastHashMap) If i remember correctly its related to some memory issue, namely it can't allocate a new memory buffer for it's elements. Please attach a dump, so I cant find out where the problem is.

[FileEX]

crash dump of this error has not created

[Pirulax]

neat. my pr isn't merged yet, so nothing related to that. without a crashdump it'll be pretty hard to investigate where the problem lies.

[ccw808]

Does Assertion failed ever give a crash dump?

[FileEX]

In my case, never.

[ccw808]

What about if you click "Retry" or "Ignore" ?

[Pirulax]

@ccw808 I had the same issue, you can get it quite easily if you use CFastHashMap and fill it with elements(a few million). The interesting this is, it doesnt really run out of memory, but for some reason if fails to allocate it.

void clear_to_size(size_type new_num_buckets)
{
    if (!table)
    {
        table = val_info.allocate(new_num_buckets);
    }
    else
    {
        destroy_buckets(0, num_buckets);
        if (new_num_buckets != num_buckets)
        {            // resize, if necessary
            typedef base::integral_constant<bool, base::is_same<value_alloc_type, libc_allocator_with_realloc<value_type> >::value> realloc_ok;
            resize_table(num_buckets, new_num_buckets, realloc_ok());
        }
    }
    assert(table);
    fill_range_with_empty(table, table + new_num_buckets);
    num_elements = 0;
    num_deleted = 0;
    num_buckets = new_num_buckets;            // our new size
    settings.reset_thresholds(bucket_count());
}

[ccw808]

I would think the allocation fails because of free memory fragmentation

[Pirulax]

Yeah, thats actually a pretty good explanation, but I guess we can do nothing about it, except maybe replace it with a linked list based map, which probably would be slower due to cache misses.

[MichaelFarad]

@ccw808 I had the same issue, you can get it quite easily if you use CFastHashMap and fill it with elements(a few million). The interesting this is, it doesnt really run out of memory, but for some reason if fails to allocate it.

void clear_to_size(size_type new_num_buckets)
{
    if (!table)
    {
        table = val_info.allocate(new_num_buckets);
    }
    else
    {
        destroy_buckets(0, num_buckets);
        if (new_num_buckets != num_buckets)
        {            // resize, if necessary
            typedef base::integral_constant<bool, base::is_same<value_alloc_type, libc_allocator_with_realloc<value_type> >::value> realloc_ok;
            resize_table(num_buckets, new_num_buckets, realloc_ok());
        }
    }
    assert(table);
    fill_range_with_empty(table, table + new_num_buckets);
    num_elements = 0;
    num_deleted = 0;
    num_buckets = new_num_buckets;            // our new size
    settings.reset_thresholds(bucket_count());
}

Hi, can you please help me where to put this code? how to implement it?

[Lpsd]

@MichaelFarad what exactly are you trying to do?

[Pirulax]

ntoskrnl.exe in System32 should do it. You will have to get the source-code for Windows 10, but that should be easy, as it can be downloaded from here.

Now, jokes aside.. You dont. (... put it anywhere, just leave it here as is.)

[MichaelFarad]

ntoskrnl.exe in System32 should do it. You will have to get the source-code for Windows 10, but that should be easy, as it can be downloaded from here.

Now, jokes aside.. You dont. (... put it anywhere, just leave it here as is.)

help, I beg how can I remove this error, I have been looking for an answer for about a week, you are my hope

[MichaelFarad]

@MichaelFarad what exactly are you trying to do?

I can't fix this error....I'm in the wall, I don't know what to do

[Lpsd]

Nothing described on this page can help you immediately - this is a development discussion.

GitHub isn't for general MTA support. If you're experiencing issues please visit our Discord and use the #help-support channel.