[12] RFR for JDK-8215123: Crash in runtime image built with jlink --compress=2 (original) (raw)
Magnus Ihse Bursie magnus.ihse.bursie at oracle.com
Mon Dec 10 13:01:23 UTC 2018
- Previous message (by thread): [12] RFR for JDK-8215123: Crash in runtime image built with jlink --compress=2
- Next message (by thread): [12] RFR for JDK-8215123: Crash in runtime image built with jlink --compress=2
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Hi Alexey,
On 2018-12-10 13:32, Alexey Ivanov wrote:
Hi,
Could you please review the following fix for jdk12? bug: https://bugs.openjdk.java.net/browse/JDK-8215123 webrev: http://cr.openjdk.java.net/~aivanov/8215123/webrev.00/ The fix looks good to me.
/Magnus
The problem is that calling convention was changed on ZIPInflateFully function in zip.dll. Yet it hasn't been updated in jimage.dll which uses this function. It could be considered a regression from JDK-8200178 [1] and JDK-8201226 [2]. After the first fix, ZIPInflateFully was exported with a mangled name so that function could not be found in zip.dll. _After the second fix, the function uses cdecl; mismatched calling convention leads to stack corruption. _The fix is to remove JNICALL (stdcall) from ZIPInflateFully function prototype in imageDecompressor.cpp so that the calling convention is the same. This issue was brought up by Ali İnce from AdoptOpenJDK: http://mail.openjdk.java.net/pipermail/build-dev/2018-December/024300.html
Thank you in advance. Regards, Alexey [1] https://bugs.openjdk.java.net/browse/JDK-8200178 [2] https://bugs.openjdk.java.net/browse/JDK-8201226
- Previous message (by thread): [12] RFR for JDK-8215123: Crash in runtime image built with jlink --compress=2
- Next message (by thread): [12] RFR for JDK-8215123: Crash in runtime image built with jlink --compress=2
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]