Re: [OpenIndiana-discuss] Nobody for whom OI/Hipster Mate on i915/KMS is working fine wants to report?
I was testing the DVD mates again and decided to install it but could not get the partition manager to work. this is in addition to add to panel omission of the volume appelet. look forward to using this in the future. Robert On 06/09/2016 16:23, Мартин Бохниг via openindiana-discuss wrote: Ok, hi Apostolos, as said already: A pity that (almost) nobody of all those others for whom it apparently works fine took the time to report, while there are always enough individuals interested in discussing "code of conduct" and worse pogroms.. Now that the only feedback anybody hears on-list about the i915 backport is "oh, doesn't work" it appears as if it is "instable crap", to quote certain well-known geniuses. Nevertheless thanks for reporting and sorry that you are experiencing problems. Now, let's nail this issue: Yes, the same pciid 42 was already attched for the old ums i915: https://searchcode.com/codesearch/view/6730577/ alias = pci8086,42 \ And yes, it should also still work with Oracle's newer driver designed for Solaris 12 that I backported to OpenSolaris. Your output below is a good beginning at best, but what's missing: *Did you still test in 32bit mode, without 32bit i915/drm bins on the Hipster/Mate DVD?* If yes, then it wonders little that you ended with what you call a "black screen", because as there is no 32bit compiled version of i915/drm included on the DVD, how would it work? Yesterday I stated already that it *is* possible to build old IA32 bins of i915/drm, and they are for download (together with the 64bit bins) here, as said: http://opensxce.org/intelkms_working_testbins/Intel_DRMxKMS_S12_to_Illumos_backport__RELEASE/01__GLOBAL/01__BINS/TAR/ They would already be part of the Illumos-gate and would by default get built for IA32 _and_ x64, if Illumos was anything related to what they claim to be (free bird and stuff). But history took different routes on scary alarming paths. So, that's why users of outdated hardware still running 32bit kernels have to perform the installation manually after the OS is installed in text mode following the steps as outlined in http://opensxce.org/intelkms_working_testbins/Intel_DRMxKMS_S12_to_Illumos_backport__RELEASE/01__GLOBAL/01__BINS/TAR/INSTALL.txt So to test them you need at least a USB stick or a growisofs or a hdd-installation (because otherwise you have a tough time copying any 32bit variant of i915/drm into the boot_archive on wrirte-protected media). You didn't send /var/log/Xorg.0.log*, You didn't send syslog You didn't send any more information, such as where does the "the screen is black" event happen. You did not say if you are now in 64bit mode off DVD, 32bit mode off DVD and if 64bit mode, why you first said you wanted to run it in 32bit mode. From your pciid I could now derive that this is: {0x8086, 0x42, CHIP_I9XX|CHIP_I965, "Intel IGDNG_D" }, \ so i965,and in https://de.wikipedia.org/wiki/Intel-900-Serie that it appears to be: P965 [17] Broadwater 90 nm Q2 2006 ICH8 / ICH8R / ICH8-DH 533 / 800 / 1066 MHz 19 W 8 GB RAM DDR2-533/667/800 1.1 keine Q965 [18] Broadwater 90 nm Q2 2006 ICH8 / ICH8R / ICH8-DH 533 / 800 / 1066 MHz 19 W 8 GB RAM DDR2-533/667/800 1.1 GMA 3000 But as you didn't tell me more about your Netbook (not even its name, only its vendor), I can only *guess* if it is 64bit capable: PROBABLY NOT. So then again tell me how you can expect it to function with the current Hipster/Mate KMS LiveDVD, after I explained long and well *that* and *why* it does *not* contain IA32 bins for i915/drm? If you want to test my backport on your vintage hardware, then you need to perform the installation in text mode or vesa mode and then take the 32bit bins after installation from the link I provided. Again: http://opensxce.org/intelkms_working_testbins/Intel_DRMxKMS_S12_to_Illumos_backport__RELEASE/01__GLOBAL/01__BINS/TAR/ When I try the live DVD on this machine I see a black screen. Did you read what I wrote yesterday already? NO 32bit bins of i915/drm on this current iso/DVD. %martin bochnig Вторник, 6 сентября 2016, 14:23 UTC от Apostolos Syropoulos: Hello, As I reported yesterday I have tested the live DVD on a system that has an Intel processor. The system is running a pre-KMS version of Xorg where the i915 is working. Here are the details: node name: display Vendor: Intel Corporation Device: Core Processor Integrated Graphics Controller Sub-Vendor: ASUSTeK Computer Inc. binding name: pci8086,42 devfs path: /pci@0,0/display@2 bus addr: 2 pci path: 0,2,0 compatible name: (pci8086,42.1043.8383.18)(pci8086,42.1043.8383)(pci1043,8383)(pci8086,42.18)(pci8086,42)(pciclass,03)(pciclass,0300) driver name:
Re: [OpenIndiana-discuss] GCC 5.4 added to the repository
Thomas Wagner писал 06.09.2016 20:53: Thank you Alex! Looks good for libgcc_s.so and libstdc++.so.6 Only it makes me wonder, why the elfdump output for none of the other (non-gcc-runtime-)libs prints "[ LAZY ]" the line right before the lib. Interesting is, that even 4.9 has none of the libs with "[ LAZY ]". As having "lazy load" for libs is there to improve startup latency for binaries/libs, I would like to ask if someone might find an example with actually printing "[ LAZY ]". This can be a completely different binary/project but compiled with enther the g++ 4.9 or the new g++ 5 version. It should print "[ LAZY ]" for at least one library. Then I would be completely happy. Best Regards, Thomas Looking at some real applications look different (g++ 4.9): $ elfdump -d /usr/bin/mate-system-monitor Dynamic Section: .dynamic index tagvalue [0] POSFLAG_1 0x1 [ LAZY ] [1] NEEDED0x2b7fa libgmodule-2.0.so.0 [2] POSFLAG_1 0x1 [ LAZY ] [3] NEEDED0x2b6f5 libpthread.so.1 [4] POSFLAG_1 0x1 [ LAZY ] [5] NEEDED0x2b80e libgtop-2.0.so.7 [6] POSFLAG_1 0x1 [ LAZY ] [7] NEEDED0x2b81f libwnck-3.so.0 [8] POSFLAG_1 0x1 [ LAZY ] [9] NEEDED0x2b82e libgtkmm-3.0.so.1 [10] POSFLAG_1 0x1 [ LAZY ] [11] NEEDED0x2b840 libgdkmm-3.0.so.1 [12] POSFLAG_1 0x1 [ LAZY ] [13] NEEDED0x2b852 libgtk-3.so.0 [14] POSFLAG_1 0x1 [ LAZY ] [15] NEEDED0x2b860 libgdk-3.so.0 [16] POSFLAG_1 0x1 [ LAZY ] [17] NEEDED0x2b86e libpangocairo-1.0.so.0 [18] POSFLAG_1 0x1 [ LAZY ] [19] NEEDED0x2b885 libpango-1.0.so.0 [20] POSFLAG_1 0x1 [ LAZY ] [21] NEEDED0x2b70e libxml2.so.2 [22] POSFLAG_1 0x1 [ LAZY ] [23] NEEDED0x2b897 librsvg-2.so.2 [24] POSFLAG_1 0x1 [ LAZY ] [25] NEEDED0x2b8a6 libgdk_pixbuf-2.0.so.0 [26] POSFLAG_1 0x1 [ LAZY ] [27] NEEDED0x2b8bd libcairo.so.2 [28] POSFLAG_1 0x1 [ LAZY ] [29] NEEDED0x2b8cb libgiomm-2.4.so.1 [30] POSFLAG_1 0x1 [ LAZY ] [31] NEEDED0x2b8dd libgio-2.0.so.0 [32] POSFLAG_1 0x1 [ LAZY ] [33] NEEDED0x2b8ed libglibmm-2.4.so.1 [34] POSFLAG_1 0x1 [ LAZY ] [35] NEEDED0x2b900 libgobject-2.0.so.0 [36] POSFLAG_1 0x1 [ LAZY ] [37] NEEDED0x2b914 libglib-2.0.so.0 [38] POSFLAG_1 0x1 [ LAZY ] [39] NEEDED0x2b925 libsigc-2.0.so.0 [40] POSFLAG_1 0x1 [ LAZY ] [41] NEEDED0x2b724 libsocket.so.1 [42] POSFLAG_1 0x1 [ LAZY ] [43] NEEDED0x2b74e libstdc++.so.6 [44] POSFLAG_1 0x1 [ LAZY ] [45] NEEDED0x2b7a1 libm.so.2 [46] POSFLAG_1 0x1 [ LAZY ] [47] NEEDED0x2b7b4 libgcc_s.so.1 [48] NEEDED0x2b7ca libc.so.1 [49] INIT 0x80cc5b0 [50] FINI 0x80cc5e0 [51] HASH 0x80566c4 [52] STRTAB0x8068050 [53] STRSZ 0x2bb36 [54] SYMTAB0x805d2d0 [55] SYMENT0x10 [56] SUNW_SYMTAB 0x805bd90 [57] SUNW_SYMSZ0xc2c0 [58] SUNW_SORTENT 0x4 [59] SUNW_SYMSORT 0x80952d8 [60] SUNW_SYMSORTSZ0x2504 [61] CHECKSUM 0x83e4 [62] VERNEED 0x8093b88 [63] VERNEEDNUM0x7 [64] PLTRELSZ 0x15b0 [65] PLTREL0x11 [66] JMPREL0x8097844 [67] REL 0x80977dc [68] RELSZ 0x1618 [69] RELENT0x8 [70] SYMINFO 0x8053b64 [71] SYMINSZ 0x2b60 [72] SYMINENT 0x4 [73] DEBUG 0 [74] FLAGS 0 0 [75] FLAGS_1 0x100
Re: [OpenIndiana-discuss] GCC 5.4 added to the repository
Some time ago I had an attempt to build illumos with clang. The main problem is inline assembly, where clang produces a bit different code. Alex > On 06 Sep 2016, at 20:51, Aurélien Larcherwrote: > > On Tue, Sep 6, 2016 at 7:32 PM, Dmitry Kozhinov wrote: >> How about LLVM/Clang? I understand that building OI with it sounds >> unrealistic, but maybe someone doing some research in this direction? > > Clang 3.8.1 is already in the repo. > >> >> >> ___ >> openindiana-discuss mailing list >> openindiana-discuss@openindiana.org >> https://openindiana.org/mailman/listinfo/openindiana-discuss > > > > -- > --- > Praise the Caffeine embeddings > > ___ > openindiana-discuss mailing list > openindiana-discuss@openindiana.org > https://openindiana.org/mailman/listinfo/openindiana-discuss ___ openindiana-discuss mailing list openindiana-discuss@openindiana.org https://openindiana.org/mailman/listinfo/openindiana-discuss
Re: [OpenIndiana-discuss] GCC 5.4 added to the repository
Thank you Alex! Looks good for libgcc_s.so and libstdc++.so.6 Only it makes me wonder, why the elfdump output for none of the other (non-gcc-runtime-)libs prints "[ LAZY ]" the line right before the lib. Interesting is, that even 4.9 has none of the libs with "[ LAZY ]". As having "lazy load" for libs is there to improve startup latency for binaries/libs, I would like to ask if someone might find an example with actually printing "[ LAZY ]". This can be a completely different binary/project but compiled with enther the g++ 4.9 or the new g++ 5 version. It should print "[ LAZY ]" for at least one library. Then I would be completely happy. Best Regards, Thomas On Tue, Sep 06, 2016 at 05:13:10PM +0300, Alexander Pyhalov wrote: > On 09/ 6/16 04:57 PM, Thomas Wagner wrote: [...] > > > > ... > > [2] somelib.so > > [3] [ LAZY ] <- delayed load/binding, libc.so can bind funcs > > earlier then libgcc_s.so > > [4] libgcc_s.so > > ... > > [5] someotherlib.so > > [6] [ LAZY ] <- delayed load/binding, libc.so can bind funcs > > earlier then libstdc++.so > > [7] libstdc++.so > > > > Hi. > > For g++ 5: > > Dynamic Section: .dynamic > index tagvalue > [0] NEEDED0x2ad libstdc++.so.6 > [1] NEEDED0x2ee libm.so.2 > [2] NEEDED0x2f8 librt.so.1 > [3] NEEDED0x303 libgcc_s.so.1 > [4] NEEDED0x2c8 libc.so.1 > [5] INIT 0x8051030 > [6] FINI 0x8051050 > [7] RUNPATH 0x311 /usr/gcc/5/lib > [8] RPATH 0x311 /usr/gcc/5/lib > [9] HASH 0x80501c0 >[10] STRTAB0x8050630 >[11] STRSZ 0x520 >[12] SYMTAB0x8050420 >[13] SYMENT0x10 >[14] SUNW_SYMTAB 0x80502e0 >[15] SUNW_SYMSZ0x350 >[16] SUNW_SORTENT 0x4 >[17] SUNW_SYMSORT 0x8050be4 >[18] SUNW_SYMSORTSZ0x50 >[19] CHECKSUM 0x5bf6 >[20] VERNEED 0x8050b50 >[21] VERNEEDNUM0x2 >[22] FINI_ARRAY0x8061318 >[23] FINI_ARRAYSZ 0x8 >[24] INIT_ARRAY0x8061320 >[25] INIT_ARRAYSZ 0x8 >[26] PLTRELSZ 0x40 >[27] PLTREL0x11 >[28] JMPREL0x8050c64 >[29] REL 0x8050c34 >[30] RELSZ 0x70 >[31] RELENT0x8 >[32] SYMINFO 0x805013c >[33] SYMINSZ 0x84 >[34] SYMINENT 0x4 >[35] DEBUG 0 >[36] FLAGS 0 0 >[37] FLAGS_1 0 0 >[38] SUNW_STRPAD 0x200 >[39] SUNW_LDMACH 0x3eEM_AMD64 >[40] PLTGOT0x8061078 > [41-51] NULL 0 > > > For g++ 4.9: > Dynamic Section: .dynamic > index tagvalue > [0] NEEDED0x2c3 libstdc++.so.6 > [1] NEEDED0x304 libm.so.2 > [2] NEEDED0x30e librt.so.1 > [3] NEEDED0x319 libgcc_s.so.1 > [4] NEEDED0x2de libc.so.1 > [5] INIT 0x80510c0 > [6] FINI 0x80510f0 > [7] HASH 0x80501c0 > [8] STRTAB0x8050640 > [9] STRSZ 0x527 >[10] SYMTAB0x8050430 >[11] SYMENT0x10 >[12] SUNW_SYMTAB 0x80502e0 >[13] SUNW_SYMSZ0x360 >[14] SUNW_SORTENT 0x4 >[15] SUNW_SYMSORT 0x8050bfc >[16] SUNW_SYMSORTSZ0x50 >[17] CHECKSUM 0xaddb >[18] VERNEED 0x8050b68 >[19] VERNEEDNUM0x2 >[20] PLTRELSZ 0x40 >[21] PLTREL0x11 >[22] JMPREL0x8050c7c >[23] REL 0x8050c4c >[24] RELSZ 0x70 >[25] RELENT0x8 >[26] SYMINFO 0x805013c >[27] SYMINSZ 0x84 >[28] SYMINENT 0x4 >[29] DEBUG 0 >[30] FLAGS 0 0 >[31] FLAGS_1 0 0 >[32] SUNW_STRPAD 0x200 >[33] SUNW_LDMACH 0x3eEM_AMD64 >[34] PLTGOT0x806111c > [35-45] NULL 0 > > I don't see any specific patch for -zinterpose... > > > -- > Best regards, > Alexander Pyhalov, > system administrator of Southern Federal
Re: [OpenIndiana-discuss] GCC 5.4 added to the repository
On Tue, Sep 6, 2016 at 7:32 PM, Dmitry Kozhinovwrote: > How about LLVM/Clang? I understand that building OI with it sounds > unrealistic, but maybe someone doing some research in this direction? Clang 3.8.1 is already in the repo. > > > ___ > openindiana-discuss mailing list > openindiana-discuss@openindiana.org > https://openindiana.org/mailman/listinfo/openindiana-discuss -- --- Praise the Caffeine embeddings ___ openindiana-discuss mailing list openindiana-discuss@openindiana.org https://openindiana.org/mailman/listinfo/openindiana-discuss
Re: [OpenIndiana-discuss] GCC 5.4 added to the repository
How about LLVM/Clang? I understand that building OI with it sounds unrealistic, but maybe someone doing some research in this direction? ___ openindiana-discuss mailing list openindiana-discuss@openindiana.org https://openindiana.org/mailman/listinfo/openindiana-discuss
Re: [OpenIndiana-discuss] Nobody for whom OI/Hipster Mate on i915/KMS is working fine wants to report?
Ok, hi Apostolos, as said already: A pity that (almost) nobody of all those others for whom it apparently works fine took the time to report, while there are always enough individuals interested in discussing "code of conduct" and worse pogroms.. Now that the only feedback anybody hears on-list about the i915 backport is "oh, doesn't work" it appears as if it is "instable crap", to quote certain well-known geniuses. Nevertheless thanks for reporting and sorry that you are experiencing problems. Now, let's nail this issue: Yes, the same pciid 42 was already attched for the old ums i915: https://searchcode.com/codesearch/view/6730577/ alias = pci8086,42 \ And yes, it should also still work with Oracle's newer driver designed for Solaris 12 that I backported to OpenSolaris. Your output below is a good beginning at best, but what's missing: *Did you still test in 32bit mode, without 32bit i915/drm bins on the Hipster/Mate DVD?* If yes, then it wonders little that you ended with what you call a "black screen", because as there is no 32bit compiled version of i915/drm included on the DVD, how would it work? Yesterday I stated already that it *is* possible to build old IA32 bins of i915/drm, and they are for download (together with the 64bit bins) here, as said: http://opensxce.org/intelkms_working_testbins/Intel_DRMxKMS_S12_to_Illumos_backport__RELEASE/01__GLOBAL/01__BINS/TAR/ They would already be part of the Illumos-gate and would by default get built for IA32 _and_ x64, if Illumos was anything related to what they claim to be (free bird and stuff). But history took different routes on scary alarming paths. So, that's why users of outdated hardware still running 32bit kernels have to perform the installation manually after the OS is installed in text mode following the steps as outlined in http://opensxce.org/intelkms_working_testbins/Intel_DRMxKMS_S12_to_Illumos_backport__RELEASE/01__GLOBAL/01__BINS/TAR/INSTALL.txt So to test them you need at least a USB stick or a growisofs or a hdd-installation (because otherwise you have a tough time copying any 32bit variant of i915/drm into the boot_archive on wrirte-protected media). You didn't send /var/log/Xorg.0.log*, You didn't send syslog You didn't send any more information, such as where does the "the screen is black" event happen. You did not say if you are now in 64bit mode off DVD, 32bit mode off DVD and if 64bit mode, why you first said you wanted to run it in 32bit mode. From your pciid I could now derive that this is: {0x8086, 0x42, CHIP_I9XX|CHIP_I965, "Intel IGDNG_D" }, \ so i965,and in https://de.wikipedia.org/wiki/Intel-900-Serie that it appears to be: P965 [17] Broadwater 90 nm Q2 2006 ICH8 / ICH8R / ICH8-DH 533 / 800 / 1066 MHz 19 W 8 GB RAM DDR2-533/667/800 1.1 keine Q965 [18] Broadwater 90 nm Q2 2006 ICH8 / ICH8R / ICH8-DH 533 / 800 / 1066 MHz 19 W 8 GB RAM DDR2-533/667/800 1.1 GMA 3000 But as you didn't tell me more about your Netbook (not even its name, only its vendor), I can only *guess* if it is 64bit capable: PROBABLY NOT. So then again tell me how you can expect it to function with the current Hipster/Mate KMS LiveDVD, after I explained long and well *that* and *why* it does *not* contain IA32 bins for i915/drm? If you want to test my backport on your vintage hardware, then you need to perform the installation in text mode or vesa mode and then take the 32bit bins after installation from the link I provided. Again: http://opensxce.org/intelkms_working_testbins/Intel_DRMxKMS_S12_to_Illumos_backport__RELEASE/01__GLOBAL/01__BINS/TAR/ > When I try the live DVD on this machine I see a black screen. Did you read what I wrote yesterday already? NO 32bit bins of i915/drm on this current iso/DVD. %martin bochnig >Вторник, 6 сентября 2016, 14:23 UTC от Apostolos Syropoulos >: > >Hello, > >As I reported yesterday I have tested the live DVD on a system that has an >Intel processor. >The system is running a pre-KMS version of Xorg where the i915 is working. >Here are the details: > >node name: display >Vendor: Intel Corporation >Device: Core Processor Integrated Graphics >Controller >Sub-Vendor: ASUSTeK Computer Inc. >binding name: pci8086,42 >devfs path: /pci@0,0/display@2 >bus addr: 2 >pci path: 0,2,0 >compatible name: >(pci8086,42.1043.8383.18)(pci8086,42.1043.8383)(pci1043,8383)(pci8086,42.18)(pci8086,42)(pciclass,03)(pciclass,0300) >driver name:i915 >instance: 0 >driver state: Attached >ddi-no-autodetach: 1 >fm-errcb-capable: TRUE >fm-ereport-capable: TRUE >primary-controller:
Re: [OpenIndiana-discuss] Nobody for whom OI/Hipster Mate on i915/KMS is working fine wants to report?
Hello, As I reported yesterday I have tested the live DVD on a system that has an Intel processor. The system is running a pre-KMS version of Xorg where the i915 is working. Here are the details: node name: display Vendor: Intel Corporation Device: Core Processor Integrated Graphics Controller Sub-Vendor: ASUSTeK Computer Inc. binding name: pci8086,42 devfs path: /pci@0,0/display@2 bus addr: 2 pci path: 0,2,0 compatible name: (pci8086,42.1043.8383.18)(pci8086,42.1043.8383)(pci1043,8383)(pci8086,42.18)(pci8086,42)(pciclass,03)(pciclass,0300) driver name:i915 instance: 0 driver state: Attached ddi-no-autodetach: 1 fm-errcb-capable: TRUE fm-ereport-capable: TRUE primary-controller: 1 ddi-kernel-ioctl: TRUE pci-msi-capid-pointer: 90 acpi-namespace: _SB_.PCI0.GFX0 assigned-addresses: 83001010 reg:1000 compatible: pci8086,42.1043.8383.18 model: VGA compatible controller power-consumption: 1 fast-back-to-back: TRUE devsel-speed: 0 interrupts: 1 max-latency:0 min-grant: 0 subsystem-vendor-id:1043 subsystem-id: 8383 device_type:display unit-address: 2 class-code: 3 revision-id:18 vendor-id: 8086 device-id: 42 When I try the live DVD on this machine I see a black screen. In the file driver_aliases that Martin supplied ther was the following entry: i915 "pci8086,42" I guess but I am not sure that the graphics hardware should work. Is there anything I can do to help resolve this problem? A.S. -- Apostolos Syropoulos Xanthi, Greece ___ openindiana-discuss mailing list openindiana-discuss@openindiana.org https://openindiana.org/mailman/listinfo/openindiana-discuss
Re: [OpenIndiana-discuss] GCC 5.4 added to the repository
On 09/ 6/16 04:57 PM, Thomas Wagner wrote: Hi Aurelien, it would be very cool if someone could send me / the list the output of "elfdump -d" on a compiled binary with C++ code. If that is not available, then regular C-code is okay too. e.g. gcc -o hello hello.c I have no test system at hand (have only older hipster on disk). What I would appreciated to see would be a configuration that has "[ LAZY ]"-load removed (only) for gcc runtime libs. Excample "good": ... [2] somelib.so [3] libgcc_s.so ... [5] someotherlib.so [6] libstdc++.so Excample "bad": ... [2] somelib.so [3] [ LAZY ] <- delayed load/binding, libc.so can bind funcs earlier then libgcc_s.so [4] libgcc_s.so ... [5] someotherlib.so [6] [ LAZY ] <- delayed load/binding, libc.so can bind funcs earlier then libstdc++.so [7] libstdc++.so Hi. For g++ 5: Dynamic Section: .dynamic index tagvalue [0] NEEDED0x2ad libstdc++.so.6 [1] NEEDED0x2ee libm.so.2 [2] NEEDED0x2f8 librt.so.1 [3] NEEDED0x303 libgcc_s.so.1 [4] NEEDED0x2c8 libc.so.1 [5] INIT 0x8051030 [6] FINI 0x8051050 [7] RUNPATH 0x311 /usr/gcc/5/lib [8] RPATH 0x311 /usr/gcc/5/lib [9] HASH 0x80501c0 [10] STRTAB0x8050630 [11] STRSZ 0x520 [12] SYMTAB0x8050420 [13] SYMENT0x10 [14] SUNW_SYMTAB 0x80502e0 [15] SUNW_SYMSZ0x350 [16] SUNW_SORTENT 0x4 [17] SUNW_SYMSORT 0x8050be4 [18] SUNW_SYMSORTSZ0x50 [19] CHECKSUM 0x5bf6 [20] VERNEED 0x8050b50 [21] VERNEEDNUM0x2 [22] FINI_ARRAY0x8061318 [23] FINI_ARRAYSZ 0x8 [24] INIT_ARRAY0x8061320 [25] INIT_ARRAYSZ 0x8 [26] PLTRELSZ 0x40 [27] PLTREL0x11 [28] JMPREL0x8050c64 [29] REL 0x8050c34 [30] RELSZ 0x70 [31] RELENT0x8 [32] SYMINFO 0x805013c [33] SYMINSZ 0x84 [34] SYMINENT 0x4 [35] DEBUG 0 [36] FLAGS 0 0 [37] FLAGS_1 0 0 [38] SUNW_STRPAD 0x200 [39] SUNW_LDMACH 0x3eEM_AMD64 [40] PLTGOT0x8061078 [41-51] NULL 0 For g++ 4.9: Dynamic Section: .dynamic index tagvalue [0] NEEDED0x2c3 libstdc++.so.6 [1] NEEDED0x304 libm.so.2 [2] NEEDED0x30e librt.so.1 [3] NEEDED0x319 libgcc_s.so.1 [4] NEEDED0x2de libc.so.1 [5] INIT 0x80510c0 [6] FINI 0x80510f0 [7] HASH 0x80501c0 [8] STRTAB0x8050640 [9] STRSZ 0x527 [10] SYMTAB0x8050430 [11] SYMENT0x10 [12] SUNW_SYMTAB 0x80502e0 [13] SUNW_SYMSZ0x360 [14] SUNW_SORTENT 0x4 [15] SUNW_SYMSORT 0x8050bfc [16] SUNW_SYMSORTSZ0x50 [17] CHECKSUM 0xaddb [18] VERNEED 0x8050b68 [19] VERNEEDNUM0x2 [20] PLTRELSZ 0x40 [21] PLTREL0x11 [22] JMPREL0x8050c7c [23] REL 0x8050c4c [24] RELSZ 0x70 [25] RELENT0x8 [26] SYMINFO 0x805013c [27] SYMINSZ 0x84 [28] SYMINENT 0x4 [29] DEBUG 0 [30] FLAGS 0 0 [31] FLAGS_1 0 0 [32] SUNW_STRPAD 0x200 [33] SUNW_LDMACH 0x3eEM_AMD64 [34] PLTGOT0x806111c [35-45] NULL 0 I don't see any specific patch for -zinterpose... -- Best regards, Alexander Pyhalov, system administrator of Southern Federal University IT department ___ openindiana-discuss mailing list openindiana-discuss@openindiana.org https://openindiana.org/mailman/listinfo/openindiana-discuss
Re: [OpenIndiana-discuss] GCC 5.4 added to the repository
Hi Aurelien, it would be very cool if someone could send me / the list the output of "elfdump -d" on a compiled binary with C++ code. If that is not available, then regular C-code is okay too. e.g. gcc -o hello hello.c I have no test system at hand (have only older hipster on disk). What I would appreciated to see would be a configuration that has "[ LAZY ]"-load removed (only) for gcc runtime libs. Excample "good": ... [2] somelib.so [3] libgcc_s.so ... [5] someotherlib.so [6] libstdc++.so Excample "bad": ... [2] somelib.so [3] [ LAZY ] <- delayed load/binding, libc.so can bind funcs earlier then libgcc_s.so [4] libgcc_s.so ... [5] someotherlib.so [6] [ LAZY ] <- delayed load/binding, libc.so can bind funcs earlier then libstdc++.so [7] libstdc++.so This removed "LAZY"-load avoids situation where uncatched C++ exceptions are falling through into into library /usr/lib/libc.so which normally can't handle them, so binary core dumps. Prominent example is "filezilla". In SFEgcc we set "-zinterpose" to have the resulting binaries link (only) the gcc runtime without the LAZY flag. Thanks much! Thomas On Tue, Sep 06, 2016 at 03:34:21PM +0200, Aur??lien Larcher wrote: > Hello, > I wrote a short post about it: > > https://www.openindiana.org/2016/09/06/gcc-5-4-now-available-as-testing-compiler/ > > Thanks go to Rich Lowe for his fixes to the linker and to Alexander > for patches and thorough testing. > > Please comment if you think about additional information which should > be provided. > Best regards, > > Aurelien > > -- > --- > Praise the Caffeine embeddings > > ___ > openindiana-discuss mailing list > openindiana-discuss@openindiana.org > https://openindiana.org/mailman/listinfo/openindiana-discuss > -- -- Thomas Wagner Service rund um UNIX(TM), Wagner Network Services, Thomas Wagner Solaris(TM), Linux(TM)Eschenweg 21, 89174 Altheim, Germany Novell(TM), Windows(TM) TEL: +49-731-9807799, FAX: +49-731-9807711 Telekommunikation, LAN, MOBILE/CELL: +49-171-6135989 Internet-Service, Elektronik EMAIL: wag...@wagner-net.com ___ openindiana-discuss mailing list openindiana-discuss@openindiana.org https://openindiana.org/mailman/listinfo/openindiana-discuss
[OpenIndiana-discuss] GCC 5.4 added to the repository
Hello, I wrote a short post about it: https://www.openindiana.org/2016/09/06/gcc-5-4-now-available-as-testing-compiler/ Thanks go to Rich Lowe for his fixes to the linker and to Alexander for patches and thorough testing. Please comment if you think about additional information which should be provided. Best regards, Aurelien -- --- Praise the Caffeine embeddings ___ openindiana-discuss mailing list openindiana-discuss@openindiana.org https://openindiana.org/mailman/listinfo/openindiana-discuss