From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from simark.ca by simark.ca with LMTP id ydUtL2mYmmr7CSgAWB0awg (envelope-from ) for ; Fri, 04 Sep 2026 06:07:37 -0400 Authentication-Results: simark.ca; dkim=pass (2048-bit key; unprotected) header.d=intel.com header.i=@intel.com header.a=rsa-sha256 header.s=Intel header.b=REv/MUBU; dkim-atps=neutral Received: by simark.ca (Postfix, from userid 112) id B97441E09B; Fri, 04 Sep 2026 06:07:37 -0400 (EDT) X-Spam-Checker-Version: SpamAssassin 4.0.1 (2024-03-25) on simark.ca X-Spam-Level: X-Spam-Status: No, score=-6.4 required=5.0 tests=ARC_SIGNED,ARC_VALID,BAYES_00, DKIMWL_WL_HIGH,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,MAILING_LIST_MULTI, RCVD_IN_DNSWL_MED autolearn=ham autolearn_force=no version=4.0.1 Received: from vm01.sourceware.org (vm01.sourceware.org [IPv6:2620:52:6:3111::32]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange x25519 server-signature ECDSA (prime256v1) server-digest SHA256) (No client certificate requested) by simark.ca (Postfix) with ESMTPS id E90261E09B for ; Fri, 04 Sep 2026 06:07:35 -0400 (EDT) Received: from vm01.sourceware.org (localhost [IPv6:::1]) by sourceware.org (Postfix) with ESMTP id 0EF0D4855E23 for ; Fri, 4 Sep 2026 10:07:35 +0000 (GMT) DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 0EF0D4855E23 Authentication-Results: sourceware.org; dkim=pass (2048-bit key, unprotected) header.d=intel.com header.i=@intel.com header.a=rsa-sha256 header.s=Intel header.b=REv/MUBU Received: from mgamail.intel.com (mgamail.intel.com [198.175.65.18]) by sourceware.org (Postfix) with ESMTPS id DC77A485E014 for ; Fri, 4 Sep 2026 10:07:06 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org DC77A485E014 Authentication-Results: sourceware.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=intel.com ARC-Filter: OpenARC Filter v1.0.0 sourceware.org DC77A485E014 Authentication-Results: sourceware.org; arc=none smtp.remote-ip=198.175.65.18 ARC-Seal: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1788516427; cv=none; b=LVy5o50dmbUF/9QTYT4tEhd1jotBvrwcwQu+XUIHBKp0W9GY/ElwQXukZcW3NI0Ot/2y2zhFRligJ89OukTfkYNcGrYHZb7zhkgqE+JS2QbMwKmJT40fKojqTkHsYhXV8fq1TCgRI8QxyjPcdXY003/DBuC6DqQO8WTxrYrmfaI= ARC-Message-Signature: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1788516427; c=relaxed/simple; bh=4eRms7Mz1MDp/uLoXccgc4RcX4PBWUFZdV2uwHK3Svo=; h=DKIM-Signature:From:To:Subject:Date:Message-ID:MIME-Version; b=HQITBlbK6X/ZUIHX+doULxI0CAruGTQXbJC9Ob0ZBGIOw2e6AVTCftb3uc1ON2vHNrKn6BA+znQj9YfwPEGlfBZ5qNqqVaQJOYtTWx5Uqntd/t8ehevuH7IHOeh71NZIydYa6agbVX7TEOYzDIGcgslHVSsbAuaLavnbIMORtII= ARC-Authentication-Results: i=1; sourceware.org; dkim=pass (2048-bit key, unprotected) header.d=intel.com header.i=@intel.com header.a=rsa-sha256 header.s=Intel header.b=REv/MUBU DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org DC77A485E014 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1788516427; x=1820052427; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=4eRms7Mz1MDp/uLoXccgc4RcX4PBWUFZdV2uwHK3Svo=; b=REv/MUBUTdFEkuHqjKFZIF6dEUVJKi76oZLIqxOfI8w3y7aF6YMc6fum eXGL1uqXq8QxWq+MwupMOD6tBqG1xKGuch7e95STu2pUFUfWAcn6kf9+G vqVH0ZCsSAY09BWjNOUdQaLNM8UyllXu5ktlD2jJCn0zpkU+E7G6utVX9 ALwHdr2UYbx40lBmOQoi81S4FEWMQnRpFytKWwu+/qnG6QMhfxGvT0OYb TyvpgtYxtqNdunDZIg50OwWvNe4IIkBwLyTcHkZVgLl/fI8rqU8FnwEH8 WoKXi/LDlvaMDuY3rsj19JFSwDCTvqPbw9BGmAkm0gVm1EtvDWA7HZMr/ w==; X-CSE-ConnectionGUID: W8SAZwptSHq0vFysMmGJ3g== X-CSE-MsgGUID: SkTt1evLTTaYpbw8WODGIQ== X-IronPort-AV: E=McAfee;i="6800,10657,11895"; a="89069623" X-IronPort-AV: E=Sophos;i="6.25,260,1779174000"; d="scan'208";a="89069623" Received: from orviesa002.jf.intel.com ([10.64.159.142]) by orvoesa110.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 04 Sep 2026 03:07:07 -0700 X-CSE-ConnectionGUID: tKEIkfITT5aNK/KTF1SXlw== X-CSE-MsgGUID: 8IAn8hWSQgi8GZKwy44gBg== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.25,260,1779174000"; d="scan'208";a="299871006" Received: from gkldtt-dev-004.igk.intel.com (HELO localhost) ([10.123.221.202]) by orviesa002-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 04 Sep 2026 03:07:06 -0700 From: Christina Joos To: gdb-patches@sourceware.org Cc: aburgess@redhat.com Subject: [PATCH 1/2] gdb: Enable OS generated corefiles on systems with Intel AMX support. Date: Fri, 4 Sep 2026 12:06:47 +0200 Message-ID: <20260904100648.1705702-2-christina.joos@intel.com> X-Mailer: git-send-email 2.53.0 In-Reply-To: <20260904100648.1705702-1-christina.joos@intel.com> References: <20260904100648.1705702-1-christina.joos@intel.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-BeenThere: gdb-patches@sourceware.org X-Mailman-Version: 2.1.30 Precedence: list List-Id: Gdb-patches mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: gdb-patches-bounces~public-inbox=simark.ca@sourceware.org This patch addresses the issue described in https://sourceware.org/bugzilla/show_bug.cgi?id=34561. On Systems with Intel AMX support the xsave size is 11008. This xsave size is not handled by gdb/i387-tdep.c:i387_guess_xsave_layout and is causing problems for corefiles generated by the linux kernel. ~~~ $ gdb GNU gdb (GDB) 19.0.50.20260814-git Copyright (C) 2026 Free Software Foundation, Inc. [...] (gdb) core core_main_SEGV [New LWP 2162880 (id 1)] [...] Core was generated by `./main'. Program terminated with signal SIGSEGV, Segmentation fault. 4 *pointer = 3; (gdb) p $ymm0 $1 = void ~~~ We should be able to print the register $ymm0 (or any other register belonging to a feature higher than SSE). For a live debug session we can print it: ~~~ Reading symbols from main... (gdb) start Temporary breakpoint 1 at 0x1131: file main.c, line 3. Starting program: /tmp/main [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1". Temporary breakpoint 1, main () at main.c:3 3 int *pointer = 0; (gdb) p $ymm0 $1 = {v16_bfloat16 = {0, 2.342e-38, 0 }, v16_half = {0, 1.5199e-05, 0 }, v8_float = { 2.34180515e-38, 0, 0, 0, 0, 0, 0, 0}, v4_double = {8.256666972292243e-317, 0, 0, 0}, v32_int8 = {0, 0, -1, 0 }, v16_int16 = {0, 255, 0 }, v8_int32 = {16711680, 0, 0, 0, 0, 0, 0, 0}, v4_int64 = { 16711680, 0, 0, 0}, v2_int128 = {16711680, 0}} [...] ~~~ Note, that this is not reproducible for corefiles generated by the gcore command, since in such corefiles the GDB target description is available and gdb/i387-tdep.c:i387_fallback_xsave_layout configures the xsave_layout based on xcr0 derived from the target description. This patch fixes this issue by handling the new xsave size in i387_guess_xsave_layout. It is necessary even though GDB does not support AMX yet, since we still pass the full xsave_size (11008) extracted from the corefile to i387_guess_xsave_layout. The patch does not fix i387_fallback_xsave_layout to configure the new xsave size. Configuring sizeof_xsave to 2696 is fine at this point, since truncating to 2696 drops only the parts GDB doesn't model yet. As a quick solution for the upcoming release and CPUs supporting Intel AMX this patch should be sufficient. Testing: I noticed that we don't have tests for corefiles for the features AVX, AVX512 and PKRU in the GDB testsuite. So this commit adds corefile tests for AVX, AVX512 and PKRU registers. It further fixes the tests gdb.base/coredump-filter.exp and gdb.arch/i386-tls-regs.exp on systems with Intel AMX support. With my patch the warning "warning: Unexpected size of section `.reg-xstate/1373786' in core file.^M" disappears and the tests are passing again. --- gdb/i387-tdep.c | 9 +++- gdb/testsuite/gdb.arch/i386-avx.c | 6 +++ gdb/testsuite/gdb.arch/i386-avx.exp | 61 ++++++++++++++++++++++---- gdb/testsuite/gdb.arch/i386-avx512.c | 6 +++ gdb/testsuite/gdb.arch/i386-avx512.exp | 56 +++++++++++++++++++++++ gdb/testsuite/gdb.arch/i386-pkru.c | 6 +++ gdb/testsuite/gdb.arch/i386-pkru.exp | 39 ++++++++++++++++ 7 files changed, 173 insertions(+), 10 deletions(-) diff --git a/gdb/i387-tdep.c b/gdb/i387-tdep.c index f22a5e02bbb..19429536537 100644 --- a/gdb/i387-tdep.c +++ b/gdb/i387-tdep.c @@ -918,9 +918,14 @@ bool i387_guess_xsave_layout (uint64_t xcr0, size_t xsave_size, x86_xsave_layout &layout) { - if (HAS_PKRU (xcr0) && xsave_size == 2696) + if (HAS_PKRU (xcr0) && (xsave_size == 2696 || xsave_size == 11008)) { - /* Intel CPUs supporting PKRU. */ + /* Intel CPUs supporting PKRU. + Note that in this case 2 possible xsave_sizes have to be handled, + since GDB does not support Intel AMX yet and CPUs that support + this feature still have a different xsave size (xsave_size == 11008). + Otherwise, we are not able to read xsave registers in OS generated + corefiles. */ layout.avx_offset = 576; layout.k_offset = 1088; layout.zmm_h_offset = 1152; diff --git a/gdb/testsuite/gdb.arch/i386-avx.c b/gdb/testsuite/gdb.arch/i386-avx.c index b6cd89e5bea..7edcabcec8c 100644 --- a/gdb/testsuite/gdb.arch/i386-avx.c +++ b/gdb/testsuite/gdb.arch/i386-avx.c @@ -18,11 +18,14 @@ along with this program. If not, see . */ #include +#include #include "nat/x86-cpuid.h" /* Align sufficient to be able to use vmovaps. */ #define ALIGN 32 +volatile int should_dump_core_p = 1; + typedef struct { _Alignas (ALIGN) float f[8]; } v8sf_t; @@ -87,6 +90,9 @@ main (int argc, char **argv) asm ("nop"); /* first breakpoint here */ + if (should_dump_core_p) /* Crash for OS corefile. */ + abort (); + asm ( "vmovaps %%ymm0, 0(%0)\n\t" "vmovaps %%ymm1, 32(%0)\n\t" diff --git a/gdb/testsuite/gdb.arch/i386-avx.exp b/gdb/testsuite/gdb.arch/i386-avx.exp index 1a786ee8a51..1740045427a 100644 --- a/gdb/testsuite/gdb.arch/i386-avx.exp +++ b/gdb/testsuite/gdb.arch/i386-avx.exp @@ -42,12 +42,26 @@ if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable \ return } +# Check reading registers after the first breakpoint for ymm int8 and +# float types. +proc test_regs_first_bp {} { + for { set r 0 } { $r < $::nr_regs } { incr r } { + gdb_test "print \$ymm$r.v8_float" \ + ".. = \\{$r, $r.125, $r.25, $r.375, $r.5, $r.625, $r.75, $r.875\\}.*" \ + "check float contents of %ymm$r" + gdb_test "print \$ymm$r.v32_int8" \ + ".. = \\{(-?${::decimal}, ){31}-?${::decimal}\\}.*" \ + "check int8 contents of %ymm$r" + } +} clean_restart ${::testfile} if {![runto_main]} { return } +gdb_test_no_output "set should_dump_core_p=0" + gdb_test "break [gdb_get_line_number "first breakpoint here"]" \ "Breakpoint .* at .*i386-avx.c.*" \ "set first breakpoint in main" @@ -59,14 +73,7 @@ if {[is_amd64_regs_target]} { set nr_regs 8 } -for { set r 0 } { $r < $nr_regs } { incr r } { - gdb_test "print \$ymm$r.v8_float" \ - ".. = \\{$r, $r.125, $r.25, $r.375, $r.5, $r.625, $r.75, $r.875\\}.*" \ - "check float contents of %ymm$r" - gdb_test "print \$ymm$r.v32_int8" \ - ".. = \\{(-?${::decimal}, ){31}-?${::decimal}\\}.*" \ - "check int8 contents of %ymm$r" -} +test_regs_first_bp for { set r 0 } { $r < $nr_regs } { incr r } { gdb_test "set var \$ymm$r.v8_float\[0\] = $r + 10" "" "set %ymm$r" @@ -100,9 +107,47 @@ with_test_prefix "force-disable xml descriptions" { return } + gdb_test_no_output "set should_dump_core_p=0" + # With qXfer:features:read disabled, we won't know anything # about YMM registers. gdb_test "print \$ymm0" " = void" gdb_test "print \$xmm0" "v4_float.*" } } + +# Restart gdb, load the corefile generated by gcore or the OS +# and check reading registers from the corefile. +proc test_corefile {core_filename} { + clean_restart $::testfile + + gdb_test "core $core_filename" "Core was generated by .*" \ + "load corefile" + + test_regs_first_bp +} + +with_test_prefix "OS generated corefile" { + set corefile [core_find $binfile] + if { $corefile eq "" } { + unsupported "unable to generate core file" + } else { + test_corefile $corefile + } +} + +with_test_prefix "gcore corefile" { + clean_restart ${::testfile} + if { ![runto_main] } { + return + } + + set line [gdb_get_line_number "first breakpoint here"] + gdb_breakpoint $line + gdb_continue_to_breakpoint "first breakpoint here" ".*$srcfile:$line.*" + + set gcorefile $binfile.gcore + if { [gdb_gcore_cmd $gcorefile "save a corefile"] } { + test_corefile $gcorefile + } +} diff --git a/gdb/testsuite/gdb.arch/i386-avx512.c b/gdb/testsuite/gdb.arch/i386-avx512.c index b1e023ce4d4..d935fab50bc 100644 --- a/gdb/testsuite/gdb.arch/i386-avx512.c +++ b/gdb/testsuite/gdb.arch/i386-avx512.c @@ -17,8 +17,11 @@ You should have received a copy of the GNU General Public License along with this program. If not, see . */ +#include #include "x86-cpuid.h" +volatile int should_dump_core_p = 1; + typedef struct { double f[8]; @@ -233,6 +236,9 @@ main (int argc, char **argv) move_zmm_data_to_reg (); asm ("nop"); /* third breakpoint here */ + if (should_dump_core_p) /* Crash for OS corefile. */ + abort (); + /* Test script incremented values, move back to array and check values. */ move_zmm_data_to_memory (); diff --git a/gdb/testsuite/gdb.arch/i386-avx512.exp b/gdb/testsuite/gdb.arch/i386-avx512.exp index c79563366dc..6984c2280e9 100644 --- a/gdb/testsuite/gdb.arch/i386-avx512.exp +++ b/gdb/testsuite/gdb.arch/i386-avx512.exp @@ -35,6 +35,8 @@ if {![runto_main]} { return } +gdb_test_no_output "set should_dump_core_p=0" + set supports_avx512 0 set test "probe AVX512 support" @@ -181,3 +183,57 @@ gdb_test "print \$zmm0.v16_int32" "= {-1, -1, -1, -1, 0 }" if { $nr_regs >= 16 } { gdb_test "print \$zmm16.v16_int32" "= {-1 }" } + +# Test reading k and zmm registers and for double and int16 types. +proc test_regs_corefile {} { + for { set r 1 } { $r < 8 } { incr r } { + gdb_test "print/x \$k$r" \ + ".. = 0x[format %x $r]2[format %x $r]1" \ + "check contents of %k$r" + } + + for { set r 0 } { $r < $::nr_regs } { incr r } { + gdb_test "print \$zmm$r.v8_double" \ + ".. = \\{$r, $r.125, $r.25, $r.375, $r.5, $r.625, $r.75, $r.875\\}.*" \ + "check double contents of %zmm$r" + gdb_test "print \$zmm$r.v32_int16" \ + ".. = \\{(-?${::decimal}, ){31}-?${::decimal}\\}.*" \ + "check int16 contents of %zmm$r" + } +} + +# Restart gdb, load the corefile generated by gcore or the OS +# and check reading registers from the corefile. +proc test_corefile {core_filename} { + clean_restart $::testfile + + gdb_test "core $core_filename" "Core was generated by .*" \ + "load corefile" + + test_regs_corefile +} + +with_test_prefix "OS generated corefile" { + set corefile [core_find $binfile] + if { $corefile eq "" } { + unsupported "unable to generate core file" + } else { + test_corefile $corefile + } +} + +with_test_prefix "gcore corefile" { + clean_restart ${::testfile} + if { ![runto_main] } { + return + } + + set line [gdb_get_line_number "third breakpoint here"] + gdb_breakpoint $line + gdb_continue_to_breakpoint "third breakpoint here" ".*$srcfile:$line.*" + + set gcorefile $binfile.gcore + if { [gdb_gcore_cmd $gcorefile "save a corefile"] } { + test_corefile $gcorefile + } +} diff --git a/gdb/testsuite/gdb.arch/i386-pkru.c b/gdb/testsuite/gdb.arch/i386-pkru.c index 9561d1e923a..e316ca90bba 100644 --- a/gdb/testsuite/gdb.arch/i386-pkru.c +++ b/gdb/testsuite/gdb.arch/i386-pkru.c @@ -18,12 +18,15 @@ along with this program. If not, see . */ #include +#include #include "x86-cpuid.h" #ifndef NOINLINE #define NOINLINE __attribute__ ((noinline)) #endif +volatile int should_dump_core_p = 1; + unsigned int have_pkru (void) NOINLINE; static inline unsigned long @@ -83,6 +86,9 @@ main (int argc, char **argv) wrpkru (wr_value); asm ("nop\n\t"); /* break here 1. */ + if (should_dump_core_p) /* Crash for OS corefile. */ + abort (); + rd_value = rdpkru (); asm ("nop\n\t"); /* break here 2. */ } diff --git a/gdb/testsuite/gdb.arch/i386-pkru.exp b/gdb/testsuite/gdb.arch/i386-pkru.exp index b9b0b99b0de..c455891b8de 100644 --- a/gdb/testsuite/gdb.arch/i386-pkru.exp +++ b/gdb/testsuite/gdb.arch/i386-pkru.exp @@ -38,6 +38,8 @@ if {![runto_main]} { return } +gdb_test_no_output "set should_dump_core_p=0" + set supports_pkru 0 set test "probe PKRU support" gdb_test_multiple "print have_pkru()" $test { @@ -108,3 +110,40 @@ gdb_test_multiple "print /x rd_value" "variable after reading pkru" { } } } + +# Restart gdb, load the corefile generated by gcore or the OS +# and check reading the pkru register from the corefile. +proc test_corefile {core_filename} { + clean_restart $::testfile + + gdb_test "core $core_filename" "Core was generated by .*" \ + "load corefile" + + gdb_test "info register pkru" ".*pkru.*$::val1.*" \ + "read pkru register" +} + +with_test_prefix "OS generated corefile" { + set corefile [core_find $binfile] + if { $corefile eq "" } { + unsupported "unable to generate core file" + } else { + test_corefile $corefile + } +} + +with_test_prefix "gcore corefile" { + clean_restart ${::testfile} + if { ![runto_main] } { + return + } + + set line [gdb_get_line_number "break here 1"] + gdb_breakpoint $line + gdb_continue_to_breakpoint "break here 1" ".*$srcfile:$line.*" + + set gcorefile $binfile.gcore + if { [gdb_gcore_cmd $gcorefile "save a corefile"] } { + test_corefile $gcorefile + } +} -- 2.53.0 ________________________________________ Intel Deutschland GmbH Registered Address: Dornacher Strasse 1, 85622 Feldkirchen, Germany Tel: +49 (89) 99143-0 www.intel.de Managing Directors: Candice Moore, Jeffrey Schneiderman, Ramachandran Sitaraman Chairperson of the Supervisory Board: Sonja Pierer Registered Seat: Munich Commercial Register B: Amtsgericht Munich HRB 186928 This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies.