[PATCH] libelf: Fix some 32bit offset/size issues that break updating 4G+ files.

[Dmitry V. Levin]

On Thu, Jun 20, 2019 at 01:10:53AM +0200, Mark Wielaard wrote:
> On Wed, 2019-06-19 at 02:04 +0200, Mark Wielaard wrote:
> > The patch also contains a testcase. But since it is necessary to create
> > and process a 4GB+ file it is guarded by some checks to make sure the
> > machine has enough disk and memory. Do these checks look reasonable?
> > They probably prevent (make the test SKIP) on all buildbot CI workers.
> 
> We discussed this a bit on irc and it was pointed out that the machine
> also really needs to have a somewhat fast disk. After a bit of
> experimenting I found the following the be a good indicator of the
> testcase being able to run in reasonable time:
> 
> diff --git a/tests/run-large-elf-file.sh b/tests/run-large-elf-file.sh
> index 3d1bdb6..d97eec9 100755
> --- a/tests/run-large-elf-file.sh
> +++ b/tests/run-large-elf-file.sh
> @@ -35,6 +35,15 @@ if test $space_available -lt 10; then
>    exit 77
>  fi
>  
> +# Make sure the disk is reasonably fast, should be able to write 100MB/s
> +fast_disk=1
> +timeout -s9 10s dd conv=fsync if=/dev/urandom of=tempfile bs=1M count=1K \
> +  || fast_disk=0; rm tempfile

Why /dev/urandom?  I suggest to use /dev/zero instead.
Also, the check itself is quite expensive, it writes 1G into tempfile,
I suggest to move it after mem_available check.

> +if test $fast_disk -eq 0; then
> +  echo "Disk not fast enough, need at least 100MB/s"

It isn't necessarily a disk, I'd say that file system is not fast enough.


-- 
ldv
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 801 bytes
Desc: not available
URL: <http://sourceware.org/pipermail/elfutils-devel/attachments/20190620/6bfb44a6/attachment.sig>