r2674 - in trunk/kernel/source/kernel-source-2.6.11-2.6.11/debian: . patches

Sven Luther luther@costa.debian.org
Fri, 11 Mar 2005 15:48:59 +0100


Author: luther
Date: 2005-03-11 15:48:58 +0100 (Fri, 11 Mar 2005)
New Revision: 2674

Modified:
   trunk/kernel/source/kernel-source-2.6.11-2.6.11/debian/changelog
   trunk/kernel/source/kernel-source-2.6.11-2.6.11/debian/patches/TODO
Log:
Updated TODO list, only non-free firmware pruning needs to be cleared.
Added the full released 2.6.10-6 changelog entry instead of the old unreleased one.


Modified: trunk/kernel/source/kernel-source-2.6.11-2.6.11/debian/changelog
===================================================================
--- trunk/kernel/source/kernel-source-2.6.11-2.6.11/debian/changelog	2005-03-11 06:35:08 UTC (rev 2673)
+++ trunk/kernel/source/kernel-source-2.6.11-2.6.11/debian/changelog	2005-03-11 14:48:58 UTC (rev 2674)
@@ -26,7 +26,7 @@
 
  -- Sven Luther <luther@debian.org>  Mon,  7 Mar 2005 14:56:32 +0100
 
-kernel-source-2.6.10 (2.6.10-6) UNRELEASED; urgency=low
+kernel-source-2.6.10 (2.6.10-6) unstable; urgency=low
 
   * Updated kernel-tree description from Martin F Krafft
     (Simon Horman)
@@ -72,9 +72,247 @@
   * sparc64-sb1500-clock-2.6.dpatch by David Miller: enable recognition 
     of the clock chip on SunBlade 1500, it won't boot otherwise.
     (Jurij Smakov).
+
+  * 088-ibmvscsi_event_struct_use_after_free.dpatch
+    The ibmvscsi driver has paths that free evt_struct, and then proceed to
+    use it.  That's clearly a no-no in SMP/threaded contexts; once an evt_struct
+    is free, something else may grab it.  So, this patch:
+    	- moves the free_event_struct() to after usage of the evt_struct
+    	- creates a single path for cleanup
+    	- calls evt_struct->done during cleanup, which is something that
+    	  should've been happening.
+    (Andres Salomon)
+
+  * 089-i386_acpi_backwards_ifdef.dpatch
+    [I386] An ACPI related printk is wrapped in an #ifdef that should be an
+    #ifndef.  Correct that (Andres Salomon). 
     
- -- Frederik Schüler <fs@gmx.net>  Fri, 25 Feb 2005 22:49:22 +0100
+  * 090-alsa_midi_emulation_chorus_reverb_swap.dpatch
+    [ALSA] seq_midi_emul.c had CHORUS_MODE and REVERB_MODE swapped in sysex().
+    This patch fixes that (Andres Salomon).
 
+  * 091-alsa_emu8000_load_fx_skip_header.patch
+    [ALSA] emu8000's load_fx() loads a userspace blob, and should be skipping
+    over the header (Andres Salomon).
+
+  * 092-net_sched_police_locate_sanity_check_input.dpatch
+    [NET] Some sanity checks are needed to ensure payloads are the same size
+    as the structures they're being copied into.  AFAICT, there's no way for a
+    malicious user to inject a payload in here (it looks like police_locate
+    stuff is called during routing changes by root); however, I can't say that
+    I'm too familiar w/ tcf stuff (Andres Salomon).
+
+  * 093-e1000_eeprom_read_off_by_one.dpatch
+    The e1000 driver's read_eeprom and write_eeprom functions allowed a bit to
+    much data to be read/written; an extra word.  Fix that (Andres Salomon).
+
+  * 094-scsi_device_set_state_missing_oldstate.dpatch
+    [SCSI] scsi_device_set_state() might be setting a device offline, w/ an
+    oldstate of BLOCK; that shouldn't be considered an error.  Add the missing
+    state transition (Andres Salomon).
+
+  * 095-jffs2_build_filesystem_memory_leak.dpatch
+    [JFFS2] Fix memory leak in jffs2_build_filesystem(), if jffs2_scan_medium
+    fails (Andres Salomon).
+
+  * 096-mtd_formatblock_zero_before_assignment.dpatch
+    [MTD] Inside NFTL_formatblock and INFTL_formatblock, the code was previously
+    assigning values to instr, then zero'ing out the values.  Instead, move the
+    assignment to after the memset (Andres Salomon).
+
+  * 097-mtd_s3c2410_nand_inithw_calc_rate_fix.dpatch
+    [MTD] s3c2410_nand_inithw() was pulling timing information from the wrong
+    place, making the timing incorrect.  This patch makes it pull the info from
+    the right place (Andres Salomon).
+
+  * 098-jffs2_do_mount_fs_init_bad_count.dpatch
+    [JFFS2] Initialize each eraseblock's bad_count to 0 in jffs2_do_mount_fs().
+    Unitialized memory sure is fun, eh?  (Andres Salomon)
+
+  * 099-jfs_commit_inode_commit_race.dpatch
+    [JFS] Fix race in jfs_commit_inode(); before actually doing the commit,
+    retest to ensure that the inode is both dirty and linked (Andres Salomon).
+
+  * 101-ppc64_hugetlb_mm_free_pgd_unlock.dpatch
+    [PPC64] In hugetlb_mm_free_pgd(), mm->page_table_lock is locked, but never
+    unlocked in the event of an error.  This patch fixes that (Andres Salomon).
+
+  * 102-cosa_sppp_channel_init_delay_attach.dpatch
+    Fix buglet in cosa's sppp_channel_init(); do not call sppp_attach() until
+    the netdev contains info that sppp_attach needs (Andres Salomon).
+
+  * 104-wan_sdla_firmware_cap_sys_rawio_addition.dpatch
+    [SECURITY] The SDLA driver only checked CAP_NET_ADMIN when doing firmware
+    uploads.  This patch adds an additional check for CAP_SYS_RAWIO, as well
+    (Andres Salomon).
+
+  * 107-xfs_finish_reclaim_always_inode.dpatch
+    [XFS] In xfs_finish_reclaim(), xfs_ireclaim() should always be called
+    (unless there's some sort of locking problem) before returning
+    (Andres Salomon).
+
+  * 108-xfs_attrmulti_by_handle_limit_mem_alloc.dpatch
+    [SECURITY] xfs_ioctl(XFS_IOC_ATTRMULTI_BY_HANDLE) calls
+    xfs_attrmulti_by_handle, which allocates memory based on user input.  This
+    patch adds a check for a max size of memory to alloc; otherwise, a user
+    can potentially DoS the system by exhausting memory.  Not sure whether root
+    is required to open the vnode device, but to be on the safe side...
+    (Andres Salomon)
+
+  * 109-binfmt_elf_loader_solar_designer_fixes.dpatch
+    [SECURITY] Fix from Solar Designer; the binfmt_elf load routines are
+    returning incorrect values, and are not strict enough in checking the
+    number of program headers (Andres Salomon).
+
+  * 110-load_module_arg_checking.dpatch
+    If the parsing of module args failed, the module could still be loaded
+    successfully.  Fix that (Andres Salomon).
+
+  * 111-security_seclvl_kconfig_dep.dpatch
+    Add a Kconfig dependency on CRYPTO for SECURITY_SECLVL (Andres Salomon).
+
+  * 112-audit_receive_skb_double_negative_return_val.dpatch
+    audit_receive_skb negates the err it receives from audit_receive_msg.  It
+    shouldn't do that (Andres Salomon).
+
+  * 114-netfilter_private_queues.dpatch
+    [NETFILTER] Amongst netfilter users, skb frag queues were shared.  This
+    could cause problems.  See
+    http://oss.sgi.com/archives/netdev/2005-01/threads.html#01036 for more
+    details (Andres Salomon).
+
+  * 115-proc_file_read_nbytes_signedness_fix.dpatch
+    [SECURITY] Heap overflow fix in /proc; WDYBTGT3-1 on
+    http://www.guninski.com/where_do_you_want_billg_to_go_today_3.html
+    No CAN# assigned yet, afaik (Andres Salomon).
+
+  * 116-n_tty_copy_from_read_buf_signedness_fixes.dpatch
+    [SECURITY] copy_from_read_buf() fix; WDYBTGT3-2 on
+    http://www.guninski.com/where_do_you_want_billg_to_go_today_3.html
+    No CAN#, yet (Andres Salomon).
+
+  * 117-reiserfs_file_64bit_size_t_fixes.dpatch
+    [SECURITY] reiserfs integer fixes; WDYBTGT3-4 on
+    http://www.guninski.com/where_do_you_want_billg_to_go_today_3.html
+    (Andres Salomon).
+
+  * 118-i2c_sis5595_setup_pci_config_return_checks.dpatch
+    [I2C] The i2c-sis5595 was forward ported from 2.4, but the calls to
+    read the pci config registers were never updated for 2.6.  As such, they
+    are incorrectly handling the results of the function calls
+    (Andres Salomon).
+
+  * 119-i2c_viapro_i2cdump_overflow.dpatch
+    [SECURITY] Fix a very hard to exploit buffer overflow in the i2c-viapro
+    driver (Andres Salomon).
+
+  * 120-openpromfs_property_read_fix.dpatch
+    Fix an oopsable condition in Openpromfs's property_read() (Andres Salomon).
+
+  * 121-cpufreq_resume_readd.dpatch
+    [CPUFREQ] Somewhere around 2.6.6, a call to cpufreq_driver->resume() was
+    accidentally dropped.  Readd it (Andres Salomon).
+
+  * 122-cpufreq_resume_readd_2.dpatch
+    [CPUFREQ] Fix a problem w/ 121-cpufreq_resume_readd.patch, where a return
+    value was not being checked correctly (Andres Salomon).
+
+  * 123-atm_get_addr_signedness_fix.dpatch
+    [SECURITY] Fix atm_get_addr()'s usage of its size arg, by making it
+    unsigned.  WDYBTGT3-3 on
+    http://www.guninski.com/where_do_you_want_billg_to_go_today_3.html
+    (Andres Salomon).
+
+  * 125-netfilter_private_queues_2.dpatch
+    [SECURITY] Add missing bits needed to make 114-netfilter_private_queues.patch
+    compile.  Patch stolen from ubuntu (mainly to keep the same ABI)
+    (Andres Salomon).
+
+  * 126-ftdi_sio_set_serial_info_baud_base_check.dpatch
+    [USB] Change ftdi_sio's set_serial_info() to do a correct check for baud_base;
+    it should be checking if baud_base<9600 if the baud_base has changed
+    (Andres Salomon).
+
+  * 129-video_cg3_screen_blanking.dpatch
+    [SPARC] Fix cg3 blanking; the driver was setting _ENABLE_VIDEO on POWERDOWN,
+    instead of unsetting it (Andres Salomon).
+
+  * 130-sparc_prom_nodematch_check_getproperty.dpatch
+    [SPARC] In prom_nodematch, check whether prom_getproperty() actually
+    succeeds before using the string it sets (Andres Salomon).
+
+  * 131-sparc_check_prom_getproperty.dpatch
+    [SPARC] Check return value from prom_getproperty() in various places where
+    it wasn't being checked (Andres Salomon).
+
+  * 132-sparc32_get_tv32_use_correct_variable.dpatch
+    [SPARC] get_tv32() uses a non-existent variable 'tv32'.  Fix that
+    (Andres Salomon).
+
+  * 133-scsi_advansys_build_with_non_pci.dpatch
+    [SCSI] Allow advansys driver to compile if CONFIG_PCI isn't set
+    (Andres Salomon).
+
+  * 134-cciss_scsi_detect_put_host_on_error.dpatch
+    [SCSI] cciss_scsi_detect() calls scsi_add_host(), which bumps the refcount
+    (even in the event of an error).  Thus, if scsi_add_host fails, the
+    scsi host refcount needs to be decremented; so, call scsi_host_put upon
+    error (Andres Salomon).
+
+  * 135-64bit_sys_shmget_compat_size_t_overflow.dpatch
+    64bit archs that offer 32bit compat wrappers for sys_shmget were mostly
+    passing the second arg as a 32bit signed int; what would happen then is,
+    it would be casted to a size_t (64bit unsigned), and the sign would cause
+    it to overflow.  Instead, we need to cast to a 32bit unsigned value first,
+    and then cast to 64bit unsigned (Andres Salomon).
+
+  * 136-64bit_sys_compat_overflows.dpatch
+    More of the same as 135*.dpatch, except for stuff like sys_ipc, sys_semget,
+    sys_msgsnd, etc (Andres Salomon).
+
+  * 137-ppc64_prom_initialize_tce_table_typo.dpatch
+    [PPC64] prom_initialize_tce_table() refers to 'vbase', which doesn't
+    actually exist; instead, 'base' was what was meant (Andres Salomon).
+
+  * 138-tulip_de_init_one_irq_init.dpatch
+    The tulip driver's de_init_one() was using pdev->irq before it had been
+    initialized.  Move its usage until after it has been initted
+    (Andres Salomon).
+
+  * 139-pci_dma_free_coherent.dpatch
+    [I386] dma_free_coherent() was calling kmalloc with its args reversed;
+    clearly incorrect (Andres Salomon).
+
+  * 140-s390_memset_arg_order_fixes.dpatch
+    [S390] Fix various drivers that call memset() with args in the wrong order
+    (Andres Salomon).
+
+  * 141-pci_devices_dont_disable_dev_if_busy.dpatch
+    For various pci devices, if pci_request_regions fails (because resources
+    are already in use), don't disable the pci device (someone else is using it)
+    (Andres Salomon).
+
+  * 142-r8169_dev_alloc_skb_alignment_fix.dpatch
+    The r8169 driver wasn't alloc'ing enough memory for skbs; the size should
+    be padded by NET_IP_ALIGN (Andres Salomon).
+
+  * 143-sysfs_write_file_signedness_problem.dpatch
+    [SYSFS] sysfs_write_file assigns the result of fill_write_buffer (which is
+    signed and returns negative upon error) to an unsigned int.  Clearly, bad
+    and wrong.. (Andres Salomon)
+
+  * 144-sys_epoll_wait_int_overflow.dpatch
+    [SECURITY] sys_epoll_wait contains an integer overflow; see
+    http://seclists.org/lists/fulldisclosure/2005/Mar/0293.html for additional
+    details (Andres Salomon).
+
+  * ipv4-fragment-queues.dpatch, ipv4-fragment-queues-2.dpatch:
+    Drop netfilter frag queue stuff, as a) it's an ABI change,
+    b) it's fixed in 2.6.11, and c) it's not that critical.  It's more
+    important to get other 2.6.10 fixes out to people (Andres Salomon).
+ 
+ -- Andres Salomon <dilinger@voxel.net>  Thu, 10 Mar 2005 18:25:39 -0500
+
 kernel-source-2.6.10 (2.6.10-5) unstable; urgency=low
 
   * Change $((exp) | exp) to $( (exp) | exp), so things work with dash

Modified: trunk/kernel/source/kernel-source-2.6.11-2.6.11/debian/patches/TODO
===================================================================
--- trunk/kernel/source/kernel-source-2.6.11-2.6.11/debian/patches/TODO	2005-03-11 06:35:08 UTC (rev 2673)
+++ trunk/kernel/source/kernel-source-2.6.11-2.6.11/debian/patches/TODO	2005-03-11 14:48:58 UTC (rev 2674)
@@ -1,9 +1,9 @@
 # + remove-references-to-removed-drivers.dpatch	# FAILED
-# + ia64-generic-no-smp.dpatch			# FAILED
-# + ia64-generic-no-smp-1-to-2.dpatch		# FAILED
 # tg3: sed script to just kill the tg3 firmware 
 #      + keep the patch to add firmware loading separate
 
 # + ipv4-fragment-queues-2.dpatch		# DROP, code seems to have dissapeared upstream ?
+# + outs.dpatch					# DROP, dropped on trippeh's recomendation.
 
-# + outs.dpatch					# DROP, dropped on trippeh's recomendation.
+# + ia64-generic-no-smp.dpatch			# DROP, a new patch is needed.
+# + ia64-generic-no-smp-1-to-2.dpatch		# DROP, a new patch is needed.