Re: [Bug binutils/4970] new library handling in binutils causes trouble with --enable-shared

[Nick Clifton]

Hi Mike,


the way to reproduce is:
 - build/install binutils like normal except add --enable-shared to configure
 - build/install binutils again with --enable-shared, and it'll fail during the
compile stages


(Just to confirm: between the first step and the second step above I should 
change my PATH so that the newly installed binaries from the first step are now 
ahead of any other binaries and so will be used in the build of the second step).


I tried this, but both builds and installs ran to completion and I was able to 
run the resulting installed binaries without problems.


One possible cause for this was that I was using the current mainline sources 
with the patch from this email applied:


  http://sourceware.org/ml/binutils/2007-10/msg00019.html

Although the patch is for a related problem I am not sure that it is 
responsible for making my builds/installs work whereas yours do not.  Could you 
put together a small shell script that I could use to reproduce the problem ?



my fix for the problem would be to not set LD_LIBRARY_PATH at all


Do you have a patch that does this, so that I could take a look at it ?

Cheers
  Nick



___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug binutils/4970] new library handling in binutils causes trouble with --enable-shared

[nickc at redhat dot com]

--- Additional Comments From nickc at redhat dot com  2007-10-02 12:16 
---
Subject: Re:  new library handling in binutils causes trouble
 with --enable-shared

Hi Mike,

> the way to reproduce is:
>  - build/install binutils like normal except add --enable-shared to configure
>  - build/install binutils again with --enable-shared, and it'll fail during 
> the
> compile stages

(Just to confirm: between the first step and the second step above I should 
change my PATH so that the newly installed binaries from the first step are now 
ahead of any other binaries and so will be used in the build of the second 
step).

I tried this, but both builds and installs ran to completion and I was able to 
run the resulting installed binaries without problems.

One possible cause for this was that I was using the current mainline sources 
with the patch from this email applied:

   http://sourceware.org/ml/binutils/2007-10/msg00019.html

Although the patch is for a related problem I am not sure that it is 
responsible for making my builds/installs work whereas yours do not.  Could you 
put together a small shell script that I could use to reproduce the problem ?

> my fix for the problem would be to not set LD_LIBRARY_PATH at all

Do you have a patch that does this, so that I could take a look at it ?

Cheers
   Nick



-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4970

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug ld/4844] auto-import and const data import problem

[nickc at redhat dot com]

--- Additional Comments From nickc at redhat dot com  2007-10-02 13:01 
---
Hi Ralf, Hi Eric,

  Ralf - thanks for the testsuite additions.  I have checked them in together
with a few formatting tidy-ups and the changelog entry below.

  Eric - I am not sure that making the .rdata section writeable is as simple as
it is for the .text section and I am not at all sure that this is something that
we want to do.  Wouldn't it be easier to have pe_find_data_imports() issue a
warning message if it has to auto import when the --enable-auto-import has not
been provided on the command line ?  (I think that it is better to not put read
only data into the .data section unless it is really necessary, thus the default
behaviour for the linker should be not to enable this workaround).  Something
like the attached patch maybe ?

Cheers
  Nick


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4844

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug ld/4844] auto-import and const data import problem

[nickc at redhat dot com]

--- Additional Comments From nickc at redhat dot com  2007-10-02 13:02 
---
oops, the changelog is:

ld/testsuite/
2007-10-02  Ralf Habecker  <[EMAIL PROTECTED]>

PR linker/4844
* ld-auto-import/auto-import.exp: Extend test to include running
the application and checking for some output.
* ld-auto-import/client.c (struct TEST): Include a variable field
with an offset.
(const_xyz): A const version of the xyz structure used to test the
initialization of constant data.


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4844

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug ld/4844] auto-import and const data import problem

[nickc at redhat dot com]

--- Additional Comments From nickc at redhat dot com  2007-10-02 13:03 
---
Created an attachment (id=2030)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=2030&action=view)
Generate a warning when auto-importing without --enable-auto-import


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4844

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug ld/4844] auto-import and const data import problem

[ebotcazou at gcc dot gnu dot org]

--- Additional Comments From ebotcazou at gcc dot gnu dot org  2007-10-02 
13:21 ---
> Eric - I am not sure that making the .rdata section writeable is as simple
> as it is for the .text section and I am not at all sure that this is
> something that we want to do.

Note that this is effectively what the patch does.

> Wouldn't it be easier to have pe_find_data_imports() issue a warning
> message if it has to auto import when the --enable-auto-import has not
> been provided on the command line ?

Easier sure, but you still end up with a broken binary. :-)

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4844

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

Re: [Bug ld/4844] auto-import and const data import problem

[Nick Clifton]

Hi Eric,


Eric - I am not sure that making the .rdata section writeable is as simple
as it is for the .text section and I am not at all sure that this is
something that we want to do.


Note that this is effectively what the patch does.


Point taken.


Wouldn't it be easier to have pe_find_data_imports() issue a warning
message if it has to auto import when the --enable-auto-import has not
been provided on the command line ?


Easier sure, but you still end up with a broken binary. :-)


But only if you are using constant data structures.  (That need to be 
auto-imported ?)  And only if the user did not put --enable-auto-import on 
their command line.


The point is, I worry about the bugs that will be caused by putting read only 
data into the read/write data section.  (Or maybe the bugs that will not be 
caught by so doing).  This fix for auto-import is a hack and I would really 
prefer it if it was not the linker's default behaviour.  So to me, providing a 
warning message and telling the user of the command line option that they ought 
to be using is an acceptable compromise.  If a programmer's app starts to seg 
fault, then one thing they ought to do is go back over the build logs and look 
for any messages from the compiler and linker telling them of potential problems.


Cheers
  Nick



___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug ld/4844] auto-import and const data import problem

[nickc at redhat dot com]

--- Additional Comments From nickc at redhat dot com  2007-10-02 14:31 
---
Subject: Re:  auto-import and const data import problem

Hi Eric,

>> Eric - I am not sure that making the .rdata section writeable is as simple
>> as it is for the .text section and I am not at all sure that this is
>> something that we want to do.
> 
> Note that this is effectively what the patch does.

Point taken.

>> Wouldn't it be easier to have pe_find_data_imports() issue a warning
>> message if it has to auto import when the --enable-auto-import has not
>> been provided on the command line ?
> 
> Easier sure, but you still end up with a broken binary. :-)

But only if you are using constant data structures.  (That need to be 
auto-imported ?)  And only if the user did not put --enable-auto-import on 
their command line.

The point is, I worry about the bugs that will be caused by putting read only 
data into the read/write data section.  (Or maybe the bugs that will not be 
caught by so doing).  This fix for auto-import is a hack and I would really 
prefer it if it was not the linker's default behaviour.  So to me, providing a 
warning message and telling the user of the command line option that they ought 
to be using is an acceptable compromise.  If a programmer's app starts to seg 
fault, then one thing they ought to do is go back over the build logs and look 
for any messages from the compiler and linker telling them of potential 
problems.

Cheers
   Nick



-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4844

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug gas/5086] New: Error message mismatch with the described action

[karvjorm at users dot sourceforge dot net]

The following error message in gas-2.17.90.pot is probably faulty:

#: config/tc-avr.c:1079 config/tc-avr.c:1095 config/tc-avr.c:1209
#: config/tc-msp430.c:2009 config/tc-msp430.c:2027
#, c-format
msgid "odd address operand: %ld"

An example how this error message is used:

if (value & 1) 
   as_bad_where (fixP->fx_file, fixP->fx_line, 
 _("odd address operand: %ld"), value);

It should be "even address operand: %ld", shouldn't it?

-- 
   Summary: Error message mismatch with the described action
   Product: binutils
   Version: 2.17
Status: NEW
  Severity: normal
  Priority: P2
 Component: gas
AssignedTo: unassigned at sources dot redhat dot com
ReportedBy: karvjorm at users dot sourceforge dot net
CC: bug-binutils at gnu dot org


http://sourceware.org/bugzilla/show_bug.cgi?id=5086

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug gas/5087] New: Error message does not report, if number value is below zero

[karvjorm at users dot sourceforge dot net]

This error message in gas-2.17.90.pot is not quite correct:

#: config/tc-avr.c:504
#, c-format
msgid "number must be less than %d"

In the source code is actually checked, if the "number" is greater than max
value or below zero.

-- 
   Summary: Error message does not report, if number value is below
zero
   Product: binutils
   Version: 2.17
Status: NEW
  Severity: normal
  Priority: P2
 Component: gas
AssignedTo: unassigned at sources dot redhat dot com
ReportedBy: karvjorm at users dot sourceforge dot net
CC: bug-binutils at gnu dot org


http://sourceware.org/bugzilla/show_bug.cgi?id=5087

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug gas/5088] New: An 8-bit range checking overlaps a constant range greater than 8-bit

[karvjorm at users dot sourceforge dot net]

The following error message is a bit limited:

#: config/tc-avr.c:530 config/tc-avr.c:665
#, c-format
msgid "constant out of 8-bit range: %d"

In the source code is actually checked, if constant is below -255 or above 255.
This range (-255...255) includes two 8-bit ranges, doesn't it?

-- 
   Summary: An 8-bit range checking overlaps a constant range
greater than 8-bit
   Product: binutils
   Version: 2.17
Status: NEW
  Severity: normal
  Priority: P2
 Component: gas
AssignedTo: unassigned at sources dot redhat dot com
ReportedBy: karvjorm at users dot sourceforge dot net
CC: bug-binutils at gnu dot org


http://sourceware.org/bugzilla/show_bug.cgi?id=5088

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug gas/5089] New: Spelling fix for duplicate

[karvjorm at users dot sourceforge dot net]

The following message in gas-2.17.90.pot should include "duplicate", not
"dupicate", I suppose.

#: config/tc-arm.c:3054
msgid "dupicate .handlerdata directive"

-- 
   Summary: Spelling fix for duplicate
   Product: binutils
   Version: 2.17
Status: NEW
  Severity: normal
  Priority: P2
 Component: gas
AssignedTo: unassigned at sources dot redhat dot com
ReportedBy: karvjorm at users dot sourceforge dot net
CC: bug-binutils at gnu dot org


http://sourceware.org/bugzilla/show_bug.cgi?id=5089

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug gas/5090] New: Spelling fix for instruction

[karvjorm at users dot sourceforge dot net]

The following message in gas-2.17.90.pot should include "instruction", not
"instrunction", shouldn't it?

#: config/tc-arm.c:14190
msgid "thumb conditional instrunction not in IT block"

-- 
   Summary: Spelling fix for instruction
   Product: binutils
   Version: 2.17
Status: NEW
  Severity: normal
  Priority: P2
 Component: gas
AssignedTo: unassigned at sources dot redhat dot com
ReportedBy: karvjorm at users dot sourceforge dot net
CC: bug-binutils at gnu dot org


http://sourceware.org/bugzilla/show_bug.cgi?id=5090

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug gas/5086] Error message mismatch with the described action

[schwab at suse dot de]

--- Additional Comments From schwab at suse dot de  2007-10-02 16:14 ---
1 is odd, not even.

-- 
   What|Removed |Added

 Status|NEW |RESOLVED
 Resolution||INVALID
Summary|Error message mismatch with |Error message mismatch with
   |the described action|the described action


http://sourceware.org/bugzilla/show_bug.cgi?id=5086

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug ld/4988] [2.18/HEAD regression] Assertion failures in ld

[jb at jblache dot org]

--- Additional Comments From jb at jblache dot org  2007-10-02 17:42 ---
Hi,

>   The test case is missing a few libraries: -llber -lsasl2 -lssl -lcrypto
>   -lexpat -lresolv -ldl.

Oops, sorry I missed those. I've put the missing libraries up at


libdl is included in the original testcase tarball, it's libdl-2.6.1.so which I
didn't rename like I did for the others. libc & other standard stuff are in the
g++-libs subdir in the original testcase, too.

Thanks,

JB.

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4988

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug binutils/4970] new library handling in binutils causes trouble with --enable-shared

[vapier at gentoo dot org]

--- Additional Comments From vapier at gentoo dot org  2007-10-02 17:45 
---
i think that thread is fleshing itself out such that it'll fix the issue i 
raised

i opened the bug because i couldnt seem to get traction on the mailing list, but
that seems to be changing with the new thread, so i'll just follow up to the
mailing list from now on until it gets resolved ...

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4970

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug ld/4844] auto-import and const data import problem

[ebotcazou at gcc dot gnu dot org]

--- Additional Comments From ebotcazou at gcc dot gnu dot org  2007-10-02 
18:19 ---
> But only if you are using constant data structures.  (That need to be 
> auto-imported ?)  And only if the user did not put --enable-auto-import on 
> their command line.

Constant data structures referencing symbols from auto-imported DLLs.  Note
that GCC doesn't downgrade them to writable memory in spite of the runtime
relocations (the mechanism to that effect is not enabled for PE-COFF).

> The point is, I worry about the bugs that will be caused by putting read
> only data into the read/write data section.  (Or maybe the bugs that will
> not be caught by so doing).  This fix for auto-import is a hack and I would
> really prefer it if it was not the linker's default behaviour.

Not very different from making the code entirely writable.  The head comment
in pe-dll.c even says that this doesn't conform to the PE-COFF specs.

> So to me, providing a warning message and telling the user of the command
> line option that they ought to be using is an acceptable compromise.  If a
> programmer's app starts to seg fault, then one thing they ought to do is go
> back over the build logs and look for any messages from the compiler and
> linker telling them of potential problems.

All right, this makes sense.


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4844

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug ld/5091] New: Mis-aligned Makefile args to genscripts.sh

[org dot sourceware at pooryorick dot com]

After configure, The following line of Makefile is missing an argument, causing
the word 'no' to be taken as the argument to native_lib_dirs instead of 
use_sysroot

GENSCRIPTS = LIB_PATH='${LIB_PATH}' $(SHELL) $(srcdir)/genscripts.sh "${srcdir}"
 "${libdir}" "${prefix}" "${exec_prefix}" i686-pc-linux-gnu i686-pc-linux-gnu 
"" "" no

-- 
   Summary: Mis-aligned Makefile args to genscripts.sh
   Product: binutils
   Version: 2.18
Status: NEW
  Severity: normal
  Priority: P2
 Component: ld
AssignedTo: unassigned at sources dot redhat dot com
ReportedBy: org dot sourceware at pooryorick dot com
CC: bug-binutils at gnu dot org
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu


http://sourceware.org/bugzilla/show_bug.cgi?id=5091

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug ld/5091] Mis-aligned Makefile args to genscripts.sh

[schwab at suse dot de]

--- Additional Comments From schwab at suse dot de  2007-10-02 21:44 ---
How did you configure it?  target_alias should never be empty.

-- 
   What|Removed |Added

 Status|NEW |WAITING


http://sourceware.org/bugzilla/show_bug.cgi?id=5091

--- You are receiving this mail because: ---
You are on the CC list for the bug, or are watching someone who is.


___
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils

[Bug binutils/5059] absolute objcopy not working on amd64?

[happyarch at gmail dot com]

--- Additional Comments From happyarch at gmail dot com  2007-10-02 23:13 
---
>I don't think they are correct. The correct ones may be

>and $0x, %esi
>and $0x, %edi

so i tried another one
root:/sources/grub-0.97# vi +1852 stage2/asm.S

//and   $0x, %rsi
and $0x, %esi

//and   $0x, %rdi
and $0x, %edi

>> root:/sources/grub-0.97# make
>...
>> boot.c: In function ‘load_image’:
>> boot.c:80: warning: cast from pointer to integer of different size
>> boot.c:80: warning: cast from pointer to integer of different size

>Those warnings have nothing to do with binutils. We can't help you here.

Those warnings means nothing, but below error messages means something.
"and $0x, %esi"  
"and $0x, %rsi"
both modification result same error message: 

/tmp/ccgHKQuT.s:795: Error: Incorrect register `%rax' used with `l' suffix
/tmp/ccgHKQuT.s:847: Error: Incorrect register `%rax' used with `l' suffix
make[2]: *** [pre_stage2_exec-disk_io.o] Error 1

is the above error messages the bug?
http://sourceware.org/bugzilla/show_bug.cgi?id=5080

TIA



root:/sources/grub-0.97# make
make  all-recursive
make[1]: Entering directory `/sources/grub-0.97'
Making all in netboot
make[2]: Entering directory `/sources/grub-0.97/netboot'
make[2]: Nothing to be done for `all'.
make[2]: Leaving directory `/sources/grub-0.97/netboot'
Making all in stage2
make[2]: Entering directory `/sources/grub-0.97/stage2'
if gcc -DHAVE_CONFIG_H -I. -I. -I.. -I../stage1  -Wall -Wmissing-prototypes
-Wunused -Wshadow -Wpointer-arith -falign-jumps=1 -falign-loops=1
-falign-functions=1 -Wundef -Os -fno-strict-aliasing -fno-stack-protector
-fno-builtin -nostdinc  -DSUPPORT_SERIAL=1 -DSUPPORT_HERCULES=1 -DFSYS_EXT2FS=1
-DFSYS_FAT=1 -DFSYS_FFS=1 -DFSYS_UFS2=1 -DFSYS_MINIX=1 -DFSYS_REISERFS=1
-DFSYS_VSTAFS=1 -DFSYS_JFS=1 -DFSYS_XFS=1 -DFSYS_ISO9660=1 -DUSE_MD5_PASSWORDS=1
-m64 -g -MT pre_stage2_exec-disk_io.o -MD -MP -MF
".deps/pre_stage2_exec-disk_io.Tpo" -c -o pre_stage2_exec-disk_io.o `test -f
'disk_io.c' || echo './'`disk_io.c; \
then mv -f ".deps/pre_stage2_exec-disk_io.Tpo"
".deps/pre_stage2_exec-disk_io.Po"; else rm -f
".deps/pre_stage2_exec-disk_io.Tpo"; exit 1; fi
disk_io.c: In function ‘make_saved_active’:
disk_io.c:473: warning: cast from pointer to integer of different size
disk_io.c:473: warning: cast to pointer from integer of different size
disk_io.c:473: warning: cast from pointer to integer of different size
disk_io.c:473: warning: cast to pointer from integer of different size
disk_io.c:473: warning: cast from pointer to integer of different size
disk_io.c:473: warning: cast to pointer from integer of different size
disk_io.c:480: warning: cast from pointer to integer of different size
disk_io.c:480: warning: cast to pointer from integer of different size
disk_io.c:486: warning: cast from pointer to integer of different size
disk_io.c:486: warning: cast to pointer from integer of different size
disk_io.c:489: warning: cast from pointer to integer of different size
disk_io.c:489: warning: cast to pointer from integer of different size
disk_io.c: In function ‘set_partition_hidden_flag’:
disk_io.c:540: warning: cast from pointer to integer of different size
disk_io.c:540: warning: cast to pointer from integer of different size
disk_io.c:542: warning: cast from pointer to integer of different size
disk_io.c:542: warning: cast to pointer from integer of different size
disk_io.c: In function ‘next_bsd_partition’:
disk_io.c:615: warning: cast from pointer to integer of different size
disk_io.c:615: warning: cast to pointer from integer of different size
disk_io.c:625: warning: cast from pointer to integer of different size
disk_io.c:625: warning: cast to pointer from integer of different size
disk_io.c:627: warning: cast from pointer to integer of different size
disk_io.c:627: warning: cast to pointer from integer of different size
disk_io.c:631: warning: cast from pointer to integer of different size
disk_io.c:631: warning: cast to pointer from integer of different size
disk_io.c:632: warning: cast from pointer to integer of different size
disk_io.c:632: warning: cast to pointer from integer of different size
disk_io.c:633: warning: cast from pointer to integer of different size
disk_io.c:633: warning: cast to pointer from integer of different size
disk_io.c:638: warning: cast from pointer to integer of different size
disk_io.c:638: warning: cast to pointer from integer of different size
disk_io.c: In function ‘next_pc_slice’:
disk_io.c:670: warning: cast from pointer to integer of different size
disk_io.c:670: warning: cast to pointer from integer of different size
disk_io.c:687: warning: cast from pointer to integer of different size
disk_io.c:687: warning: cast to pointer from integer of different size
disk_io.c:687: warning: cast from pointer to integer of different size
disk_io.c:687: warn