RFR: 8241439: jdk.NativeLibraryEvent hooks all opened regular files
Yasumasa Suenaga
suenaga at oss.nttdata.com
Thu Apr 9 11:55:14 UTC 2020
Hi Denghui,
On 2020/04/09 19:39, Denghui Dong wrote:
> Hi Yasumasa,
>
> I have another question:
> if a module is loaded into several discontinuous segments, I think this patch can't cover this situation, am I right?
I think all executables would load into continuous memory spaces in Linux.
Linux would load ELF binary at load_elf_binary(). This function loads all loadable segments (PT_LOAD)
and start / end addresses are stored to mm_struct [1].
If the application attempt to load same library in twice with dlopen(3), manpage [2] says same handle is returned.
So the library would not deploy to discontinuous space.
Thanks,
Yasumasa
[1] https://github.com/torvalds/linux/blob/master/fs/binfmt_elf.c#L1103
[2] https://linux.die.net/man/3/dlopen
> Denghui
>
> ------------------------------------------------------------------
> From:Yasumasa Suenaga <suenaga at oss.nttdata.com>
> Send Time:2020年4月9日(星期四) 17:05
> To:董登辉(卓昂) <denghui.ddh at alibaba-inc.com>; Hotspot dev runtime <hotspot-runtime-dev at openjdk.java.net>; hotspot-jfr-dev at openjdk.java.net <hotspot-jfr-dev at openjdk.java.net>; hotspot-jfr-dev <hotspot-jfr-dev-bounces at openjdk.java.net>
> Cc:yasuenag at gmail.com <yasuenag at gmail.com>
> Subject:Re: RFR: 8241439: jdk.NativeLibraryEvent hooks all opened regular files
>
> Hi Denghui,
>
> Thank you for the comment.
> I uploaded new webrev:
>
> http://cr.openjdk.java.net/~ysuenaga/JDK-8241439/webrev.01/
>
> On 2020/04/09 17:11, Denghui Dong wrote:
> > Hi Yasumasa,
> > I'm not a reviewer, but the following code looks strange:
> > =====
> > 2096 if ((base == NULL) && (base < phdr_base)) {
> > =====
> >
> > if the condition is right, it could be simplified: if (base < phdr_base),
> > but IMO, the correct condition should be if (base > phdr_base), I may be wrong.
>
> You are right, condition should be (base > phdr_base). We need to store base address to `base` from all PT_LOAD header.
> So NULL check is needed because (base > phdr_base) always false without it.
>
>
> Thanks,
>
> Yasumasa
>
>
> > Thanks,
> > Denghui
> > ------------------------------------------------------------------
> > From:Yasumasa Suenaga <suenaga at oss.nttdata.com>
> > Send Time:2020年4月9日(星期四) 15:01
> > To:Hotspot dev runtime <hotspot-runtime-dev at openjdk.java.net>; hotspot-jfr-dev at openjdk.java.net <hotspot-jfr-dev at openjdk.java.net>
> > Cc:yasuenag at gmail.com <yasuenag at gmail.com>
> > Subject:RFR: 8241439: jdk.NativeLibraryEvent hooks all opened regular files
> >
> > Hi all,
> >
> > Please review this change:
> >
> > JBS: https://bugs.openjdk.java.net/browse/JDK-8241439
> > webrev: http://cr.openjdk.java.net/~ysuenaga/JDK-8241439/webrev.00/
> >
> > HotSpot has os::get_loaded_modules_info() to get all loaded native libraries, however it would call callback in twice on same library on Linux. Also it would call callback even if the file is not a library.
> >
> > We can see this problem in jdk.NativeLibrary JFR event as below:
> >
> >
> > ```
> > jdk.NativeLibrary {
> > startTime = 16:59:06.621
> > name = "/usr/lib64/libc-2.30.so"
> > baseAddress = 0x7FDB06ACC000
> > topAddress = 0x7FDB06AF1000
> > }
> >
> > jdk.NativeLibrary {
> > startTime = 16:59:06.621
> > name = "/usr/lib64/libc-2.30.so"
> > baseAddress = 0x7FDB06AF1000
> > topAddress = 0x7FDB06C40000
> > }
> >
> > :
> >
> > ```
> >
> > ```
> > jdk.NativeLibrary {
> > startTime = 16:59:06.621
> > name = "/home/ysuenaga/OpenJDK/jdk/build/linux-x86_64-server-fastdebug/images/jdk/lib/server/classes.jsa"
> > baseAddress = 0x800000000
> > topAddress = 0x800007000
> > }
> > ```
> >
> > This change uses dl_iterate_phdr() to get them, so these problem has gone.
> >
> >
> > Thanks,
> >
> > Yasumasa
> >
>
>
More information about the hotspot-runtime-dev
mailing list