On 16-12-2015 10:40, Xinze Chi (信泽) wrote:
Because we use the new strategy for filejournal in master branch. the
write entry submit to writeq is aligned already.
So if assert at this line, this strategy should have bug.
I do not know why you have some heads #include, Maybe you modify the
logic in filejournal?
No, the adds I've done are to work around the fact that the
linux_version stuff
is not really going to work for FreeBSD. Not the test, nor the resulting
code.
So the header part of the file now looks like:
#include "common/blkdev.h"
#if defined(__linux__)
#include "common/linux_version.h"
#endif
#if defined(__FreeBSD__)
#include "common/freebsd_version.h"
#define O_DSYNC O_SYNC
#endif
The remainder of the diffs are about locking when using aio, which i do
not ATM.
So never say never with software, but I think I've left the logic as it
is/was.
--WjW
2015-12-16 17:20 GMT+08:00 Willem Jan Withagen <wjw@xxxxxxxxxxx>:
On 16-12-2015 02:57, Xinze Chi (信泽) wrote:
You mean your ceph assert(0 == "bl should be align"), right?
But in master branch, the 1036 line is not assert(0 == "bl should be align").
Yes you are correct. I think I have some heade #includes why this moves
down in my file.
None the less I still get trapped on that specific assert.
Next question is of course why this code is what it is. Since once the
assert triggers, the remainder does not get executed.
Unless compiled with NDEBUG, then only the warning gets printed.
But the other asserts never get triggered.
So back to my original question, Why have this very stringent test and
than in test/buffer.cc forgo the fact that this could/should be aligned.
--WjW
2015-12-16 7:56 GMT+08:00 Willem Jan Withagen <wjw@xxxxxxxxxxx>:
Hi,
I'm receiving traps when running the tests going with 'gmake check'
and on one of the tests it traps on:
os/FileJournal.cc:1036
void FileJournal::align_bl(off64_t pos, bufferlist& bl)
{
// make sure list segments are page aligned
if (directio && (!bl.is_aligned(block_size) ||
!bl.is_n_align_sized(CEPH_MINIMUM_BLOCK_SIZE))) {
assert(0 == "bl should be align");
if ((bl.length() & (CEPH_MINIMUM_BLOCK_SIZE - 1)) != 0 ||
(pos & (CEPH_MINIMUM_BLOCK_SIZE - 1)) != 0)
dout(0) << "rebuild_page_aligned failed, " << bl << dendl;
assert((bl.length() & (CEPH_MINIMUM_BLOCK_SIZE - 1)) == 0);
assert((pos & (CEPH_MINIMUM_BLOCK_SIZE - 1)) == 0);
}
}
And then I get confused with the following commit in other tests:
commit 8ed724222651812c2ee8cc3804dc1f54c973897d
Author: Kefu Chai <kchai@xxxxxxxxxx>
Date: Fri Sep 4 01:23:31 2015 +0800
test/bufferlist: do not expect !is_page_aligned() after unaligned
rebuild
if the size of a bufferlist is page aligned we allocate page aligned
memory chunk for it when rebuild() is called. otherwise we just call
the plain new() to allocate new memory chunk for holding the continuous
buffer. but we should not expect that `new` allocator always returns
unaligned memory chunks. instead, it *could* return page aligned
memory chunk as long as the allocator feels appropriate. so, the
`EXPECT_FALSE(bl.is_page_aligned())` after the `rebuild()` call is
removed.
Signed-off-by: Kefu Chai <kchai@xxxxxxxxxx>
Could these 2 be related, and do I have an alignment problem when
allocating buffers and bufferlists....
Note that I also have not solved the illegal writes to _len in
bufferlists when running unittest_erasure_code_shec_arguments.
So any suggestions as to where to look at for this, are welcome.
--WjW
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
