Re: [osv-dev] Demand paging for malloc
I mean "allocated but not used" case.
In the example, it only uses the first 4KB of 512MB. 4KB is a small portion
of 128MB.
I also totally agree with mmap-backed malloc for large memory allocation.
Wonsup Yoon
2019년 9월 10일 화요일 오후 3시 52분 52초 UTC+9, Nadav Har'El 님의 말:
>
>
> On Mon, Sep 9, 2019 at 2:41 PM > wrote:
>
>> Hi,
>> I found malloc returns physical address in mempool area and does not
>> perform demand paging (only mmap does).
>> Is there any reason for the design choice?
>>
>
> I guess you're not really asking about *demand* paging ("swapping")
> because this feature is usually an unnecessary complication in
> single-application kernels. If I understand correctly, your question more
> about why does malloc() allocate physically contiguous memory unlike mmap().
>
> The answer is that we originally did this because of the issue of huge
> pages. Modern CPUs have another level above the regular 4K pages - 2 MB
> pages called "huge pages". Applications get a performance boost by using
> huge pages, because the CPU's page table cache (the TLB) can only fit a
> fixed number of pages, so an application using few huge tables instead of a
> large number of small pages will have a higher hit rate in this cache, and
> improved performance. So it is inefficient to allocate a 8 KB allocation
> using small pages (two separate pages in physical pages but contiguous in
> virtual memory) - it is more efficient to set up huge pages and return the
> 8KB allocation as a contiguous part of such a huge-table. We measured this
> to noticeably improve (by a few percent) of applications which use memory
> allocated in small and-medium sized allocations.
>
> That being said, for really large allocations - significantly over 2MB
> (the huge-page size) - there's no real reason why we need those to be
> contiguous in physical memory - we can build them from 2MB huge-pages, each
> contiguous in physical memory but overall the entire object is not. In
> fact, this is *exactly* what our mmap() does. So it would be nice if
> malloc() could fall back to call mmap() for allocations larger than some
> threshold (2MB, 4MB, or whatever). This is definitely doable - we have an
> open issue about this: https://github.com/cloudius-systems/osv/issues/854
> - and it explains how it can be done.
>
>
>> OSv fails, even if it only uses small portion of allocated memory.
>>
>
> In your example, if I understand correctly, you tried to allocate 512 MB
> with a 128 MB memory, so it's not "a small portion" of memory - it's more
> than the memory you have :-)
>
> But the issue still has merit. If you tried to allocate 50 MB it might
> have still have failed, because of memory fragmentation (i.e., we have 50
> MB free memory, but not contiguous in physical memory).
>
>
>>
>> #include
>> #include
>> #include
>>
>> int main()
>> {
>> size_t size = 512 * 1024 * 1024;
>> printf("Hello from main\n");
>> printf("allocation %x start\n", size);
>> //int *p = (int *)malloc(size); // FAIL
>> int *p = (int *)mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_PRIVATE |
>> MAP_ANONYMOUS, -1, 0); // OK
>> printf("allocation %x = %p\n", size, p);
>> *(p) = 512;
>> printf("access done\n");
>>
>> return 0;
>> }
>>
>>
>> Thanks.
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "OSv Development" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to osv...@googlegroups.com .
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/osv-dev/5378fb86-73b9-4987-aa0a-70a573d1921b%40googlegroups.com
>>
>>
>> .
>>
>
--
You received this message because you are subscribed to the Google Groups "OSv
Development" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to osv-dev+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/osv-dev/be6f748f-e472-4705-8d94-e16e44061c7e%40googlegroups.com.
[osv-dev] Re: Demand paging for malloc
It only fails on firecracker whose default memory size is 128M.
For QEMU, if I set VM's memory to 128M, it also fails.
./scripts/build -j24 image=native-example && ./scripts/run.py -m 128M
trace:
OSv v0.53.0-87-gf7b6bee5
eth0: 192.168.122.15
Booted up in 345.76 ms
Hello from main
allocation 2000 start
Unreasonable allocation attempt, larger than memory. Aborting.
[backtrace]
0x403e29b4
0x403e5da1
0x403e60b7
0x403e63f6
0x1140094f
0x4042d39c
0x4020dfe3
0x4042d568
0x40461be5
0x403f9fb6
0x40399e52
0x9f01e98d66991fff
0x403f997f
0x4156415741e58947
Wonsup
2019년 9월 10일 화요일 오전 12시 35분 7초 UTC+9, Waldek Kozaczuk 님의 말:
>
> Interesting. I cannot reproduce the malloc() problem. I have no issues
> running your example with uncommented malloc:
>
> #include
> #include
> #include
>
> int main()
> {
> size_t size = 512 * 1024 * 1024;
> printf("Hello from main\n");
> printf("allocation %lx start\n", size);
> int *p = (int *)malloc(size); // FAIL
> //int *p = (int *)mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_PRIVATE |
> MAP_ANONYMOUS, -1, 0); // OK
> printf("allocation %lx = %p\n", size, p);
> *(p) = 512;
> printf("access done\n");
>
> return 0;
> }
>
> OSv v0.53.0-107-g4f59f284
>
> eth0: 192.168.122.15
>
> Booted up in 135.90 ms
>
> Cmdline: /test_large --help
>
> Hello from main
>
> allocation 2000 start
>
> allocation 2000 = 0x81b0d040
>
> access done
>
> As you can see the address returned is a virtual one. Can you email your
> thread stack trace when it crashes in your case?
>
> As far as why we commit the memory for all malloc I can not answer this
> and others might be able to address why it was designed like this.
>
> This might be a relevant issue -
> https://github.com/cloudius-systems/osv/issues/854. I guess we might
> change implementation of malloc_large() and for large sizes create a VMA
> like for mmap. But then malloc is used all over the place in kernel code so
> I wonder if we would not have issues like nested faults - see this -
> https://github.com/cloudius-systems/osv/issues/143 - it has some
> background between malloc() and mmap() handling in OSv.
>
> Waldek
>
> On Monday, September 9, 2019 at 7:41:31 AM UTC-4, pusno...@gmail.com
> wrote:
>>
>> Hi,
>> I found malloc returns physical address in mempool area and does not
>> perform demand paging (only mmap does).
>> Is there any reason for the design choice?
>> OSv fails, even if it only uses small portion of allocated memory.
>>
>>
>> #include
>> #include
>> #include
>>
>> int main()
>> {
>> size_t size = 512 * 1024 * 1024;
>> printf("Hello from main\n");
>> printf("allocation %x start\n", size);
>> //int *p = (int *)malloc(size); // FAIL
>> int *p = (int *)mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_PRIVATE |
>> MAP_ANONYMOUS, -1, 0); // OK
>> printf("allocation %x = %p\n", size, p);
>> *(p) = 512;
>> printf("access done\n");
>>
>> return 0;
>> }
>>
>>
>> Thanks.
>>
>
--
You received this message because you are subscribed to the Google Groups "OSv
Development" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to osv-dev+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/osv-dev/27528a92-b136-4f85-a914-647d55656613%40googlegroups.com.
[osv-dev] Re: Memory layout
Great, Thanks! 2019년 9월 5일 목요일 오후 12시 58분 56초 UTC+9, Waldek Kozaczuk 님의 말: > > There is no official Wiki but details of this issue should give you an > idea - https://github.com/cloudius-systems/osv/issues/1043 also this > commit that has changed kernel placement - > https://github.com/cloudius-systems/osv/commit/2a1795db8a22b0b963a64d068f5d8acc93e5785d > . > > There is also this email about memory allocation - > https://groups.google.com/d/msg/osv-dev/Xw3nGuPWzMw/tYVj1d-GAQAJ. Please > note that many improvements have been since then. > > Lastly, we are looking for volunteers. So if you have a bit of time to > write a Wiki page about it, you are welcome to do so. > > Waldek > > On Wednesday, September 4, 2019 at 6:41:41 AM UTC-4, pusno...@gmail.com > wrote: >> >> Hi, I'm working around OSv especially for memory management. >> >> I'm curious that is there any "documented" memory layout information such >> as address ranges and limitations. >> >> It seems that there are various ranges (mmap (annon) and malloc do not >> alloc same address ranges) and many limitations (eg. stack pointers do not >> go through virtio (virt_to_phys fails)). >> >> Thanks! >> >> -- You received this message because you are subscribed to the Google Groups "OSv Development" group. To unsubscribe from this group and stop receiving emails from it, send an email to osv-dev+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/osv-dev/c3ba4c05-90b2-4dff-b59d-9cf972aee6fb%40googlegroups.com.
