Issue tracker: Issues
https://tracker.crystax.net/
https://tracker.crystax.net/favicon.ico?1535920027
2015-01-26T00:10:53Z
Issue tracker
Redmine
CrystaX NDK - Task #885 (Open): Implement own function to be able fetch "getprop" values from system
https://tracker.crystax.net/issues/885
2015-01-26T00:10:53Z
Dmitry M.
dm@crystax.net
<p>Previously, there was __system_property_get() function provided by Bionic. In latest Android versions, it was removed from platform headers, but still remains in Bionic. Currently we've implemented thin dlsym-ed wrapper for it in libcrystax. However, no guarantee that it will persist in Bionic in the future, so we need implement own analogue.</p>
CrystaX NDK - Bug #830 (Open): long double print formatting broken on x86_64 when being built by ...
https://tracker.crystax.net/issues/830
2014-12-28T12:19:46Z
Dmitry M.
dm@crystax.net
<p>RUN test on emulator-5554 [x86_64]: crystax-test-stdio2 (test-printfloat)<br>
/opt/android/android-sdk-mac/platform-tools/adb -s "emulator-5554" shell "cd /data/local/tmp/ndk-tests && LD_LIBRARY_PATH=/data/local/tmp/ndk-tests ./test-printfloat"<br>
1..11<br>
jni/test-printfloat.c:69: printf("%13LE", 1.0L) ==> [2.983231E-4947], expected [ 1.000000E+00]<br>
jni/test-printfloat.c:75: printf("%13Lf", 1.0L) ==> [ 0.000000], expected [ 1.000000]<br>
jni/test-printfloat.c:76: printf("%13LG", 1.0L) ==> [2.98323E-4947], expected [ 1]<br>
Aborted<br>
134<br>
---> TEST FAILED!!</p>
CrystaX NDK - Bug #829 (Open): sscanf(" 0X.0123456789abcdefffp-3g", "%Le%s", &ld, buf) return wr...
https://tracker.crystax.net/issues/829
2014-12-28T12:13:34Z
Dmitry M.
dm@crystax.net
CrystaX NDK - Bug #828 (Open): sscanf("3.14159265358979323846", "%Lg", &ld) return wrong value on...
https://tracker.crystax.net/issues/828
2014-12-28T12:09:22Z
Dmitry M.
dm@crystax.net
<p>RUN test on emulator-5554 [x86_64]: crystax-test-stdio2 (test-scanfloat)<br>
/opt/android/android-sdk-mac/platform-tools/adb -s "emulator-5554" shell "cd /data/local/tmp/ndk-tests && LD_LIBRARY_PATH=/data/local/tmp/ndk-tests ./test-scanfloat"<br>
1..4<br>
jni/test-scanfloat.c:169: ERROR: Assertion failed: "eq(LDBL, ld, 3.14159265358979323846L)"<br>
Aborted<br>
134<br>
---> TEST FAILED!!</p>
CrystaX NDK - Bug #827 (Open): sscanf("+1.234568e-52", "%LG", &ld) return wrong value on x86_64 w...
https://tracker.crystax.net/issues/827
2014-12-28T12:05:10Z
Dmitry M.
dm@crystax.net
<p>RUN test on emulator-5554 [x86_64]: crystax-test-stdio2 (test-scanfloat)<br>
/opt/android/android-sdk-mac/platform-tools/adb -s "emulator-5554" shell "cd /data/local/tmp/ndk-tests && LD_LIBRARY_PATH=/data/local/tmp/ndk-tests ./test-scanfloat"<br>
1..4<br>
jni/test-scanfloat.c:104: ERROR: Assertion failed: "eq(LDBL, ld, 1.234568e-52L)"<br>
Aborted<br>
134<br>
---> TEST FAILED!!</p>
CrystaX NDK - Bug #826 (Open): sscanf("NaN", "%e", &f) return non-NaN on arm and x86_64 when bein...
https://tracker.crystax.net/issues/826
2014-12-28T11:39:13Z
Dmitry M.
dm@crystax.net
<p>RUN test on Galaxy Nexus [armeabi-v7a]: crystax-test-stdio2 (test-scanfloat)<br>
/opt/android/android-sdk-mac/platform-tools/adb -s "014E04DE10004008" shell "cd /data/local/tmp/ndk-tests && LD_LIBRARY_PATH=/data/local/tmp/ndk-tests ./test-scanfloat"<br>
1..4<br>
ok 1 - scanfloat<br>
jni/test-scanfloat.c:228: ERROR: Assertion failed: "fetestexcept(FE_INVALID) == 0"<br>
Aborted<br>
134<br>
---> TEST FAILED!!</p>
CrystaX NDK - Bug #821 (Open): sscanf("nan(1234)", "%e", &f) return non-NaN on mips device
https://tracker.crystax.net/issues/821
2014-12-26T17:58:51Z
Dmitry M.
dm@crystax.net
<p>Running device test [mips]: crystax-test-stdio2 (test-scanfloat)<br>
/opt/android/android-sdk-mac/platform-tools/adb -s "Novo7 Paladin" shell "cd /data/local/tmp/ndk-tests && LD_LIBRARY_PATH=/data/local/tmp/ndk-tests ./test-scanfloat"<br>
1..4<br>
ok 1 - scanfloat<br>
assertion "fetestexcept(FE_INVALID) == 0" failed: file "jni/test-scanfloat.c", line 201, function "main"<br>
Bus error<br>
138<br>
---> TEST FAILED!!</p>
CrystaX NDK - Bug #820 (Open): Hex rounding of floating point values works wrong for ARM soft float
https://tracker.crystax.net/issues/820
2014-12-26T17:00:49Z
Dmitry M.
dm@crystax.net
<p>Running device test [armeabi]: crystax-test-stdio2 (test-printfloat)<br>
/opt/android/android-sdk-mac/platform-tools/adb -s "HT011P800177" shell "cd /data/local/tmp/ndk-tests && LD_LIBRARY_PATH=/data/local/tmp/ndk-tests ./test-printfloat"<br>
1..11<br>
ok 1 - printfloat<br>
ok 2 - printfloat<br>
ok 3 - printfloat<br>
ok 4 - printfloat<br>
ok 5 - printfloat<br>
ok 6 - printfloat<br>
ok 7 - printfloat<br>
ok 8 - printfloat<br>
ok 9 - printfloat<br>
ok 10 - printfloat<br>
jni/test-printfloat.c:326: printf("%.11A", 0x1.23456789abcdep0) ==> [0X1.23456789ABDP+0], expected [0X1.23456789ABCP+0]<br>
1<br>
---> TEST FAILED!!</p>
<p>It only happens for 'armeabi' binaries, i.e. built in soft float mode</p>
CrystaX NDK - Bug #819 (Open): Decimal rounding of floating point values don't take into account ...
https://tracker.crystax.net/issues/819
2014-12-26T16:53:22Z
Dmitry M.
dm@crystax.net
<p>No matter which rounding mode set with @fesetround@ (@FE_DOWNWARD@, @FE_UPWARD@, @FE_TOWARDZERO@ or @FE_TONEAREST@), @printf("%.3f", flt)@ and @printf("%.3Lf", flt)@ always produce results like if @FE_TONEAREST@ was set. It happens for all platforms, no matter soft/hard float or 32/64-bit signedness. The same happens with scanf() family of functions.</p>
CrystaX NDK - Bug #818 (Open): printf(FMT, [+/-]HUGE_VALL) produce wrong result on 32-bit platforms
https://tracker.crystax.net/issues/818
2014-12-26T15:15:20Z
Dmitry M.
dm@crystax.net
<p>Running device test [armeabi-v7a]: crystax-test-stdio2 (test-printfloat)<br>
/opt/android/android-sdk-mac/platform-tools/adb -s "HT011P800177" shell "cd /data/local/tmp/ndk-tests && LD_LIBRARY_PATH=/data/local/tmp/ndk-tests ./test-printfloat"<br>
1..11<br>
ok 1 - printfloat<br>
jni/test-printfloat.c:106: printf("%4.2Le", HUGE_VALL) ==> [1.80e+308], expected [ inf]<br>
1<br>
---> TEST FAILED!!</p>
<p>Running device test [armeabi-v7a]: crystax-test-stdio2 (test-printfloat)<br>
/opt/android/android-sdk-mac/platform-tools/adb -s "HT011P800177" shell "cd /data/local/tmp/ndk-tests && LD_LIBRARY_PATH=/data/local/tmp/ndk-tests ./test-printfloat"<br>
1..11<br>
ok 1 - printfloat<br>
jni/test-printfloat.c:112: printf("%Lf", -HUGE_VALL) ==> [-17976931348623159077293051907890247336179769789423065727343008115773267580550096313270847732240753], expected [-inf]<br>
1<br>
---> TEST FAILED!!</p>
CrystaX NDK - Bug #815 (Open): printf("%tu", (size_t)-1) produce wrong string on 32-bit platforms
https://tracker.crystax.net/issues/815
2014-12-26T14:22:17Z
Dmitry M.
dm@crystax.net
<p>It produce "18446744073709551615" (UINT64_MAX) instead of "4294967295" (UINT32_MAX)</p>
CrystaX NDK - Bug #770 (Open): tests/device/crazy_linker failed
https://tracker.crystax.net/issues/770
2014-12-23T01:11:55Z
Dmitry M.
dm@crystax.net
<p>*** TESTS [gcc4.9]: 5 of 16 FAILED<br>
[armeabi-v7a] tests/device/crazy_linker (test_dl_wrappers)<br>
[armeabi-v7a] tests/device/crazy_linker (test_load_library)<br>
[armeabi-v7a] tests/device/crazy_linker (test_load_library_callbacks)<br>
[armeabi-v7a] tests/device/crazy_linker (test_load_library_depends)<br>
[armeabi-v7a] tests/device/crazy_linker (test_two_shared_relros)</p>
<p>Please note the only failed sub-tests are those which load dynamically another shared library. Here is call stack from one of such crash:<br>
<br>
F/libc (22596): Fatal signal 11 (SIGSEGV), code 1, fault addr 0x1c in tid 22596 (test_dl_wrapper)<br>
I/DEBUG ( 1511): *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***<br>
I/DEBUG ( 1511): Build fingerprint: 'google/hammerhead/hammerhead:5.0.1/LRX22C/1602158:user/release-keys'<br>
I/DEBUG ( 1511): Revision: '11'<br>
I/DEBUG ( 1511): ABI: 'arm'<br>
I/DEBUG ( 1511): pid: 22596, tid: 22596, name: test_dl_wrapper >>> ./test_dl_wrappers <<<<br>
I/DEBUG ( 1511): signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x1c<br>
I/DEBUG ( 1511): r0 0000001c r1 b6dd6cb8 r2 42206d6f r3 bef9ba24<br>
I/DEBUG ( 1511): r4 0000001c r5 b6ddad4c r6 b6b2b140 r7 b6f42955<br>
I/DEBUG ( 1511): r8 00000000 r9 00000000 sl 00000000 fp bef9ba9c<br>
I/DEBUG ( 1511): ip b6f1c68c sp bef9ba08 lr b6ee662d pc b6ecadb6 cpsr 000f0030<br>
I/DEBUG ( 1511):<br>
I/DEBUG ( 1511): backtrace:<br>
I/DEBUG ( 1511): #00 pc 00016db6 /system/lib/libc.so (pthread_mutex_lock+7)<br>
I/DEBUG ( 1511): #01 pc 00032629 /system/lib/libc.so (fputs+26)<br>
I/DEBUG ( 1511): #02 pc 00000d13 /data/local/tmp/ndk-tests/libbar.so</p>
<p>The same picture for other sub-tests. It's easy to see libbar.so call 'fputs' from libc.so instead of libcrystax.so, even though libbar.so was linked with -lcrystax before -lc. This looks like bug in crazy_linker which don't take care about order of symbols when resolving them on library load.</p>
CrystaX NDK - Feature #722 (Open): Add libraw library
https://tracker.crystax.net/issues/722
2014-10-10T14:09:35Z
Dmitry M.
dm@crystax.net
CrystaX NDK - Bug #696 (Open): [tests] Build of tests/build/flto failed when built by clang
https://tracker.crystax.net/issues/696
2014-10-02T23:22:25Z
Dmitry M.
dm@crystax.net
CrystaX NDK - Task #82 (Open): Implement plugin system for VFS in libcrystax
https://tracker.crystax.net/issues/82
2013-01-08T17:21:37Z
Dmitry M.
dm@crystax.net