Le 16.02.2008 09:25, Andrew Morton a écrit :
> ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.25-rc2/2.6.25-rc2-mm1/

Got this in dmesg output:

------------[ cut here ]------------
WARNING: at arch/x86/mm/ioremap.c:129 __ioremap+0xc7/0x182()
Modules linked in:
Pid: 1, comm: swapper Not tainted 2.6.25-rc2-mm1 #40
 [<c0118955>] warn_on_slowpath+0x41/0x6d
 [<c0131d0a>] ? trace_hardirqs_on+0xb/0xd
 [<c01fdd89>] ? acpi_os_release_object+0x8/0xc
 [<c02188d4>] ? acpi_ut_delete_object_desc+0x39/0x3e
 [<c0217f05>] ? acpi_ut_delete_internal_obj+0x2c1/0x2c9
 [<c0131d0a>] ? trace_hardirqs_on+0xb/0xd
 [<c0131cde>] ? trace_hardirqs_on_caller+0xdf/0x100
 [<c0131d0a>] ? trace_hardirqs_on+0xb/0xd
 [<c01129c5>] __ioremap+0xc7/0x182
 [<c0112a99>] ioremap_nocache+0xa/0xc
 [<c02a6877>] acpi_os_map_memory+0x11/0x1a
 [<c020b697>] acpi_ex_system_memory_space_handler+0xd3/0x228
 [<c0203bd8>] ? acpi_ev_address_space_dispatch+0x142/0x1a8
 [<c020b5c4>] ? acpi_ex_system_memory_space_handler+0x0/0x228
 [<c0203bfd>] acpi_ev_address_space_dispatch+0x167/0x1a8
 [<c02083dd>] acpi_ex_access_region+0x1e4/0x270
 [<c02085bc>] acpi_ex_field_datum_io+0x153/0x2a1
 [<c0158ac0>] ? cache_alloc_debugcheck_after+0xe9/0x165
 [<c020879b>] acpi_ex_extract_from_field+0x91/0x224
 [<c0206bcf>] ? acpi_ex_read_data_from_field+0x163/0x1b0
 [<c0206bec>] acpi_ex_read_data_from_field+0x180/0x1b0
 [<c020d256>] acpi_ex_resolve_node_to_value+0x1aa/0x230
 [<c0207a32>] acpi_ex_resolve_to_value+0x270/0x2aa
 [<c0209e47>] acpi_ex_resolve_operands+0x24e/0x52f
 [<c0200827>] acpi_ds_exec_end_op+0xb7/0x4f4
 [<c0212d51>] acpi_ps_parse_loop+0x5e5/0x79c
 [<c02120dc>] acpi_ps_parse_aml+0xb2/0x2dd
 [<c021350c>] acpi_ps_execute_method+0x13d/0x20d
 [<c020fb72>] acpi_ns_evaluate+0x10e/0x1b0
 [<c02164ca>] acpi_ut_evaluate_object+0x57/0x1a1
 [<c02166ce>] acpi_ut_execute_STA+0x22/0x7b
 [<c0218d61>] ? acpi_ut_release_mutex+0x85/0x8f
 [<c020f45d>] acpi_ns_get_device_callback+0x5a/0x121
 [<c021173e>] acpi_ns_walk_namespace+0xfa/0x114
 [<c020f381>] acpi_get_devices+0x47/0x5d
 [<c020f403>] ? acpi_ns_get_device_callback+0x0/0x121
 [<c021ce4a>] ? ec_parse_device+0x0/0x6e
 [<c03a4460>] acpi_ec_ecdt_probe+0xaa/0x10a
 [<c03a404b>] acpi_init+0x73/0x21e
 [<c023e5bb>] ? class_create+0x4b/0x64
 [<c0390652>] kernel_init+0xa3/0x1f3
 [<c0103a4e>] ? restore_nocheck_notrace+0x0/0xe
 [<c03905af>] ? kernel_init+0x0/0x1f3
 [<c03905af>] ? kernel_init+0x0/0x1f3
 [<c01045a7>] kernel_thread_helper+0x7/0x10
 =======================
---[ end trace 4eaa2a86a8e2da22 ]---

The offending code in arch/x86/mm/ioremap.c is:
101 static void __iomem *__ioremap(unsigned long phys_addr, unsigned long size,
102                                enum ioremap_mode mode)
103 {
...
119         /*
120          * Don't allow anybody to remap normal RAM that we're using..
121          */
122         for (pfn = phys_addr >> PAGE_SHIFT; pfn < max_pfn_mapped &&
123              (pfn << PAGE_SHIFT) < last_addr; pfn++) {
124                 if (page_is_ram(pfn) && pfn_valid(pfn) &&
125                     !PageReserved(pfn_to_page(pfn)))
126                         return NULL;
127         }
128 
129         WARN_ON_ONCE(page_is_ram(pfn));
130 

The WARN_ON was introduced by git-agpgart.patch.

CC'd Stuart Bennett and Arjan van dev Ven.

attached: full dmesg and config.
~~
laurent