Skip to contentSkip to conversation

zlib: updated to v.1.2.3

PATCH New Alessandro Rubini 5 participants 15 messages Jul 27, 2009 – Jul 29, 2009 1 message not in the archive

lore
Share & tools
lore
#1 lore raw

PATCH v3 1/1 zlib: updated to v.1.2.3

In reply to a message not in the archive
Quoted 2 lines

Please, is there somebody that can test this patch on a differen arch than PPC?

Arm: success.

   tornado% git log --pretty=oneline | head -1
   b201171f2b4d509f3ad510b214bee70ff902e3d6 zlib: updated to v.1.2.3

U-Boot 2009.06-00494-gb201171 (Jul 27 2009 - 17:05:15)

[...]
## Booting kernel from Legacy Image at 01308000 ...

      Image Name:   Linux-2.6.31-rc3
      Image Type:   ARM Linux Kernel Image (gzip compressed)
      Data Size:    1706717 Bytes =  1.6 MB
      Load Address: 00008000
      Entry Point:  00008000
      Uncompressing Kernel Image ...

OK

/alessandro

#2 lore raw

PATCH v3 1/1 zlib: updated to v.1.2.3

Quoted 2 lines · On Monday 27 July 2009 17:09:28 Alessandro Rubini wrote:
Quoted 2 lines

Please, is there somebody that can test this patch on a differen arch than PPC?

I just noticed a common difference in the failing uncompression's. Load address is 0x0 in this case. Perhaps you could give this a try on your system as well (if possible).

I'll try to uncompress to something != 0 tomorrow on PPC.

Best regards,
Stefan

=====================================================================
DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: +49-8142-66989-0 Fax: +49-8142-66989-80 Email: office at denx.de
=====================================================================

#3 lore raw

PATCH 1/1 zlib: allow 0 as destination pointer

The entry point of an image can be 0 (like on PowerPC), so allow next_out to be NULL in inflate()

---

Stefan Roese:

Quoted 1 line

I'll try to uncompress to something != 0 tomorrow on PPC.

You are right. I naively thought my arm has RAM at 0 like PPC, but entry point of image is not really 0.

 lib_generic/zlib.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
lib_generic/zlib.c +1 -1
index f415f6b..49fb145 100644
@@ -1365,7 +1365,7 @@ int flush;
     static const unsigned short order[19] = /* permutation of code lengths */
         {16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15};
 
-    if (strm == Z_NULL || strm->state == Z_NULL || strm->next_out == Z_NULL ||
+    if (strm == Z_NULL || strm->state == Z_NULL ||
         (strm->next_in == Z_NULL && strm->avail_in != 0))
         return Z_STREAM_ERROR;
 
Signature · 1 line
1.5.6.5
#4 lore raw

PATCH 1/1 zlib: allow 0 as destination pointer

Quoted 2 lines · Alessandro Rubini

The entry point of an image can be 0 (like on PowerPC), so allow next_out to be NULL in inflate()

Hi,

this patch solves the inflate() error but now a watchdog reset is triggered during decompression:

## Booting kernel from Legacy Image at 00200000 ...

   Image Name:   Linux-2.4.37.2-dbox2
   Image Type:   PowerPC Linux Kernel Image (gzip compressed)
   Data Size:    772042 Bytes = 753.9 kB

Load Address: 00000000
Entry Point: 00000000
Verifying Checksum ... OK
Uncompressing Kernel Image ...
debug: DDF: Calibrating delay loop...
debug: DDF: 66.76 BogoMIPS
debug: WATCHDOG RESET
[...]

The neither the old nor the new code incorporate WATCHDOG_RESET() so the new code seems to be slower than the old one which works. I have yet to dig through the zlib code to insert some WATCHDOG_RESET() here and there like I did with the LZMA code.

Cheers, rhabarber1848

#5 lore raw

PATCH 1/1 zlib: allow 0 as destination pointer

Dear rhabarber1848,

Quoted 3 lines · In message <[email protected]> you wrote:

this patch solves the inflate() error but now a watchdog reset is triggered during decompression:

...

Quoted 1 line

The neither the old nor the new code incorporate WATCHDOG_RESET() so the new

You are wrong. The urrent code installs WATCHDOG_RESET as s.outcb() callback; see function zunzip() in "lib_generic/gunzip.c".

Quoted 3 lines

code seems to be slower than the old one which works. I have yet to dig through the zlib code to insert some WATCHDOG_RESET() here and there like I did with the LZMA code.

I think this is the wrong thing to do. Use the provided callback instead.

I wonder why you do not create a patch between the old original version and what wd have in U-Boot now (which should clearly show what has been changed to adapt this code for U-Boot), and then apply (if necessary, manually) that patch to the current version?

Best regards,

Wolfgang Denk

Signature · 7 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
"You ain't experienced..." "Well, nor are you." "That's true. But the
point is ... the point is ... the point is we've been not experienced
for a lot longer than you. We've got  a  lot  of  experience  of  not
having any experience."           - Terry Pratchett, _Witches Abroad_
#6 lore raw

PATCH 1/1 zlib: allow 0 as destination pointer

Quoted 4 lines · Wolfgang Denk

I wonder why you do not create a patch between the old original version and what wd have in U-Boot now (which should clearly show what has been changed to adapt this code for U-Boot), and then apply (if necessary, manually) that patch to the current version?

Hi Wolfgang,

I did exactly that, I took rc1, downloaded the zlib-1.2.3 patch from the newsgroup[1] and added the patch by Alessandro[2]. The WATCHDOG_RESET code in gunzip.c is not touched by either patches, but I still get a WATCHDOG_RESET when the kernel image is decompressed with the new code.

Though is seems that

if (z->outcb != Z_NULL)

     (*z->outcb)(Z_NULL, 0);

is not used anymore in zlib.c so the WATCHDOG_RESET calls are silently ignored. I will try to re-add them tomorrow.

Besides this, rc1 is working fine here.

Cheers, rhabarber1848

[1] http://article.gmane.org/gmane.comp.boot-loaders.u-boot/64727
[2] http://article.gmane.org/gmane.comp.boot-loaders.u-boot/64969

#7 lore raw

PATCH 1/1 zlib: allow 0 as destination pointer

Dear rhabarber1848,

Quoted 8 lines · In message <[email protected]> you wrote:

Though is seems that

if (z->outcb != Z_NULL)

     (*z->outcb)(Z_NULL, 0);

is not used anymore in zlib.c so the WATCHDOG_RESET calls are silently ignored. I will try to re-add them tomorrow.

I see.

Quoted 1 line

Besides this, rc1 is working fine here.

Thanks for the confirmation.

Best regards,

Wolfgang Denk

Signature · 7 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
"Pardon me for breathing, which I never do anyway so I don't know why
I bother to say it, oh God, I'm so depressed. Here's another of those
self-satisfied doors. Life! Don't talk to me about life."
                                        - Marvin the Paranoid Android
#8 lore raw

PATCH 1/1 zlib: allow 0 as destination pointer

In reply to Alessandro Rubini
Quoted 11 lines · On Monday 27 July 2009 18:40:31 Alessandro Rubini wrote:

The entry point of an image can be 0 (like on PowerPC), so allow next_out to be NULL in inflate()

---

Stefan Roese:

Quoted 1 line

I'll try to uncompress to something != 0 tomorrow on PPC.

You are right. I naively thought my arm has RAM at 0 like PPC, but entry point of image is not really 0.

Yes, this works:

Thanks.

Best regards,
Stefan

=====================================================================
DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: +49-8142-66989-0 Fax: +49-8142-66989-80 Email: office at denx.de
=====================================================================

#9 lore raw

PATCH 1/1 zlib: allow 0 as destination pointer

Very good!!!
So is it time to re-apply zlib patch to u-boot main tree, together Alessandro's patch?

Wolfgang, what's your position?

Best Regards,
Giuseppe

-----Original Message-----
From: [email protected] [mailto:u-boot-bounces at lists.denx.de] On Behalf Of Stefan Roese
Sent: Tuesday, July 28, 2009 8:31 AM
To: Alessandro Rubini
Cc: u-boot at lists.denx.de; rhabarber1848 at web.de
Subject: Re: [U-Boot] [PATCH] zlib: allow 0 as destination pointer

Quoted 11 lines · On Monday 27 July 2009 18:40:31 Alessandro Rubini wrote:

The entry point of an image can be 0 (like on PowerPC), so allow next_out to be NULL in inflate()

---

Stefan Roese:

Quoted 1 line

I'll try to uncompress to something != 0 tomorrow on PPC.

You are right. I naively thought my arm has RAM at 0 like PPC, but entry point of image is not really 0.

Yes, this works:

Thanks.

Best regards,
Stefan

=====================================================================
DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: +49-8142-66989-0 Fax: +49-8142-66989-80 Email: office at denx.de
=====================================================================
_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
http://lists.denx.de/mailman/listinfo/u-boot

#10 lore raw

PATCH 1/1 zlib: allow 0 as destination pointer

Dear Giuseppe CONDORELLI,

Quoted 5 lines · In message <[email protected]> you wrote:

Very good!!!
So is it time to re-apply zlib patch to u-boot main tree, together Alessandro's patch?

Wolfgang, what's your position?

Actually I'm waiting for a new patch that integrates that fix.

Best regards,

Wolfgang Denk

Signature · 5 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Just about every computer on the market today runs Unix,  except  the
Mac (and nobody cares about it).                   - Bill Joy 6/21/85
#11 lore raw

PATCH 1/1 zlib: allow 0 as destination pointer

Dear Wolfgang,

Quoted 1 line

Actually I'm waiting for a new patch that integrates that fix.

Maybe I've not understood.
Do you mean you are waiting a new zlib patch [v4] that includes Alessandro's patch, to apply
only one patch containing all?

Best Regards,
Giuseppe
-----Original Message-----
From: Wolfgang Denk [mailto:wd at denx.de]
Sent: Wednesday, July 29, 2009 8:24 AM
To: Giuseppe CONDORELLI
Cc: u-boot at lists.denx.de
Subject: Re: [U-Boot] [PATCH] zlib: allow 0 as destination pointer

Dear Giuseppe CONDORELLI,

Quoted 5 lines · In message <[email protected]> you wrote:

Very good!!!
So is it time to re-apply zlib patch to u-boot main tree, together Alessandro's patch?

Wolfgang, what's your position?

Actually I'm waiting for a new patch that integrates that fix.

Best regards,

Wolfgang Denk

Signature · 5 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Just about every computer on the market today runs Unix,  except  the
Mac (and nobody cares about it).                   - Bill Joy 6/21/85
#13 lore raw

PATCH 1/1 zlib: allow 0 as destination pointer

In reply to Wolfgang Denk
Quoted 10 lines · Wolfgang Denk

Dear Giuseppe CONDORELLI,

Quoted 5 lines · In message <[email protected]> you wrote:

Very good!!!
So is it time to re-apply zlib patch to u-boot main tree, together Alessandro's patch?

Wolfgang, what's your position?

Actually I'm waiting for a new patch that integrates that fix.

Hi,

please do not commit the zlib-1.2.3 patch in its current state, although Alessandro's patch fixes on bug there is still the problem that the WATCHDOG_RESET() calls from gunzip.c

#if defined(CONFIG_HW_WATCHDOG) || defined(CONFIG_WATCHDOG)
        s.outcb = (cb_func)WATCHDOG_RESET;
#else
        s.outcb = Z_NULL;

#endif<>/* CONFIG_HW_WATCHDOG */

are silently ignored in the new zlib code. My C knowledge is insufficient to solve this problem. In its current state it will break zlib decompression on the machines I am taking care of.

Cheers, rhabarber1848

#12 lore raw

PATCH 1/1 zlib: allow 0 as destination pointer

In reply to Giuseppe CONDORELLI

Dear Giuseppe CONDORELLI,

Quoted 6 lines · In message <[email protected]> you wrote:
Quoted 2 lines

Actually I'm waiting for a new patch that integrates that fix.

Maybe I've not understood.
Do you mean you are waiting a new zlib patch [v4] that includes Alessandro's patch, to apply
only one patch containing all?

Yes. It makes no sense to add a known to be broken patch and fix it in a second commit, as this leaves only a state in the tree where attempts to bisect any issues run into problems. So please submit a new, fixed patch, that really works.

Best regards,

Wolfgang Denk

Signature · 5 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Applying computer technology is simply finding the  right  wrench  to
pound in the correct screw.
#14 lore raw

PATCH 1/1 zlib: allow 0 as destination pointer

In reply to rhabarber1848

Dear rhabarber1848,

Quoted 11 lines · In message <[email protected]> you wrote:
Quoted 2 lines

Actually I'm waiting for a new patch that integrates that fix.

please do not commit the zlib-1.2.3 patch in its current state, although Alessandro's patch fixes on bug there is still the problem that the WATCHDOG_RESET() calls from gunzip.c

#if defined(CONFIG_HW_WATCHDOG) || defined(CONFIG_WATCHDOG)
        s.outcb = (cb_func)WATCHDOG_RESET;
#else
        s.outcb = Z_NULL;
Quoted 5 lines

#endif<>/* CONFIG_HW_WATCHDOG */

are silently ignored in the new zlib code. My C knowledge is insufficient to solve this problem. In its current state it will break zlib decompression on the machines I am taking care of.

Agreed - that should be working, too, before we pull that patch into mainline.

Best regards,

Wolfgang Denk

Signature · 6 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Perfection is reached, not when there is no longer anything  to  add,
but when there is no longer anything to take away.
                                           - Antoine de Saint-Exupery
#15 lore raw

PATCH 1/1 zlib: allow 0 as destination pointer

Hi rhabarber1848,

please could you test latest two zlib patches I have sent few minutes ago?

NOTE: the first one is waiting for approval due to its big size.

Thanks, I hope these patches will solve your issue.

G.

-----Original Message-----
From: [email protected] [mailto:u-boot-bounces at lists.denx.de] On Behalf Of Wolfgang Denk
Sent: Wednesday, July 29, 2009 9:12 AM
To: rhabarber1848 at web.de
Cc: u-boot at lists.denx.de
Subject: Re: [U-Boot] [PATCH] zlib: allow 0 as destination pointer

Dear rhabarber1848,

Quoted 7 lines · In message <[email protected]> you wrote:
Quoted 2 lines

Actually I'm waiting for a new patch that integrates that fix.

please do not commit the zlib-1.2.3 patch in its current state, although Alessandro's patch fixes on bug there is still the problem that the WATCHDOG_RESET() calls from gunzip.c

Quoted 7 lines
#if defined(CONFIG_HW_WATCHDOG) || defined(CONFIG_WATCHDOG)
        s.outcb = (cb_func)WATCHDOG_RESET;
#else
        s.outcb = Z_NULL;

#endif<>/* CONFIG_HW_WATCHDOG */

are silently ignored in the new zlib code. My C knowledge is insufficient

to

Quoted 2 lines

solve this problem. In its current state it will break zlib decompression on the machines I am taking care of.

Agreed - that should be working, too, before we pull that patch into mainline.

Best regards,

Wolfgang Denk

Signature · 10 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Perfection is reached, not when there is no longer anything  to  add,
but when there is no longer anything to take away.
                                           - Antoine de Saint-Exupery
_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
http://lists.denx.de/mailman/listinfo/u-boot