[bug#63008] [PATCH core-updates] gnu: gdb@11: Update to 11.2.

[bug#63008] [PATCH core-updates] gnu: gdb@11: Update to 11.2.

[Pierre Langlois]

[-- Attachment #1.1: Type: text/plain, Size: 341 bytes --]

Hi Guix!

I'm trying out the core-updates branch on aarch64, and gdb@11 doesn't
build with the new glibc, see: https://sourceware.org/bugzilla/show_bug.cgi?id=28302

Luckily, the patch was backported to 11.2, so it's a trivial fix! Is
this OK to commit? It'll be a big rebuild, I guess it could be bundled
with other fixes.

Thanks,
Pierre


[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 519 bytes --]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-gnu-gdb-11-Update-to-11.2.patch --]
[-- Type: text/x-patch, Size: 1195 bytes --]

From dd8ab2ac3659057241707f05a5a63910aa1da677 Mon Sep 17 00:00:00 2001
Message-Id: <dd8ab2ac3659057241707f05a5a63910aa1da677.1682115065.git.pierre.langlois@gmx.com>
From: Pierre Langlois <pierre.langlois@gmx.com>
Date: Fri, 21 Apr 2023 22:53:11 +0100
Subject: [PATCH] gnu: gdb@11: Update to 11.2.

* gnu/packages/gdb.scm (gdb-11): Update to 11.2.
---
 gnu/packages/gdb.scm | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/gnu/packages/gdb.scm b/gnu/packages/gdb.scm
index 82050d9ac5..dfc32ea2fb 100644
--- a/gnu/packages/gdb.scm
+++ b/gnu/packages/gdb.scm
@@ -48,14 +48,14 @@ (define-module (gnu packages gdb)
 (define-public gdb-11
   (package
     (name "gdb")
-    (version "11.1")
+    (version "11.2")
     (source (origin
               (method url-fetch)
               (uri (string-append "mirror://gnu/gdb/gdb-"
                                   version ".tar.xz"))
               (sha256
                (base32
-                "151z6d0265hv9cgx9zqqa4bd6vbp20hrljhd6bxl7lr0gd0crkyc"))))
+                "0cif2n3wfg1w8vc1kfnp6358idxa1zj0wjm8m5qqc6w8f5mc75ql"))))
     (build-system gnu-build-system)
     (outputs '("out" "debug"))
     (arguments
-- 
2.39.2

[bug#63008] gdb on aarch64

[Andreas Enge]

Hello Pierre,

we also have gdb@12 in core-updates; does this also not build? Right now
none of the two have been built on aarch64 on CI, so I lack an overview.

On x86_64, both build; @12 has 101 dependents, @11 7563!
This cannot be accomodated on core-updates any more, I am afraid.

We will need to build it out afterwards. And probably the best course
of action would be to move the dependents to gdb@12; or drop the
dependency? Why do so many packages depend on a debugger?

Andreas

[bug#63008] gdb on aarch64

[Pierre Langlois]

[-- Attachment #1: Type: text/plain, Size: 1346 bytes --]

Hi Andreas, sorry I missed your reply!

Andreas Enge <andreas@enge.fr> writes:

> Hello Pierre,
>
> we also have gdb@12 in core-updates; does this also not build? Right now
> none of the two have been built on aarch64 on CI, so I lack an overview.

Yup, gdb@12 builds just fine for me!

>
> On x86_64, both build; @12 has 101 dependents, @11 7563!
> This cannot be accomodated on core-updates any more, I am afraid.
>
> We will need to build it out afterwards. And probably the best course
> of action would be to move the dependents to gdb@12; or drop the
> dependency? Why do so many packages depend on a debugger?

Yeah gdb is used as a dependency for testing rust, I'm not sure if it
has to be 11, it can probably work with 12 although I haven't tried.

I think the main reason we still have gdb 11 is to make sure we don't
rebuild the rust world. I'm afraid if we don't do update it know, we'll
have to do it quite soon, can the rust world be rebuilt on a staging
branch soon after the core-updates merge? I'm concerned that these days
a lot relies on rust (via librsvg IIRC), so not having it available
might be an issue, even for simple systems :-/.

All that being said, it might be possible to add a separate gdb@11.2
package, and only use it for rust on non-x86, in a way that doesn't
cause a rebuild, could that work?

Thanks,
Pierre


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 519 bytes --]

[bug#63008] gdb on aarch64

[Pierre Langlois]

[-- Attachment #1.1: Type: text/plain, Size: 1577 bytes --]


Pierre Langlois <pierre.langlois@gmx.com> writes:

> [[PGP Signed Part:Undecided]]
> Hi Andreas, sorry I missed your reply!
>
> Andreas Enge <andreas@enge.fr> writes:
>
>> Hello Pierre,
>>
>> we also have gdb@12 in core-updates; does this also not build? Right now
>> none of the two have been built on aarch64 on CI, so I lack an overview.
>
> Yup, gdb@12 builds just fine for me!
>
>>
>> On x86_64, both build; @12 has 101 dependents, @11 7563!
>> This cannot be accomodated on core-updates any more, I am afraid.
>>
>> We will need to build it out afterwards. And probably the best course
>> of action would be to move the dependents to gdb@12; or drop the
>> dependency? Why do so many packages depend on a debugger?
>
> Yeah gdb is used as a dependency for testing rust, I'm not sure if it
> has to be 11, it can probably work with 12 although I haven't tried.
>
> I think the main reason we still have gdb 11 is to make sure we don't
> rebuild the rust world. I'm afraid if we don't do update it know, we'll
> have to do it quite soon, can the rust world be rebuilt on a staging
> branch soon after the core-updates merge? I'm concerned that these days
> a lot relies on rust (via librsvg IIRC), so not having it available
> might be an issue, even for simple systems :-/.
>
> All that being said, it might be possible to add a separate gdb@11.2
> package, and only use it for rust on non-x86, in a way that doesn't
> cause a rebuild, could that work?

To illustrate what I mean, here's a WIP patch (it needs comments),
AFAICT, this would only be a rebuild on non-x86:


[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 519 bytes --]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: gdb.patch --]
[-- Type: text/x-patch, Size: 1739 bytes --]

diff --git a/gnu/packages/gdb.scm b/gnu/packages/gdb.scm
index 82050d9ac5..d460f3950c 100644
--- a/gnu/packages/gdb.scm
+++ b/gnu/packages/gdb.scm
@@ -146,6 +146,18 @@ (define-public gdb-11
     ;; RPC stubs).
     (supported-systems (fold delete %supported-systems %hurd-systems))))
 
+(define-public gdb-11.2
+  (package
+    (inherit gdb-11)
+    (version "11.2")
+    (source (origin
+              (method url-fetch)
+              (uri (string-append "mirror://gnu/gdb/gdb-"
+                                  version ".tar.xz"))
+              (sha256
+               (base32
+                "xvczsqcbh5y0gx7qrclpna0qzx26sk7lra6y8qzxam1biyzr65qf"))))))
+
 (define-public gdb-12
   (package
     (inherit gdb-11)
@@ -166,6 +178,11 @@ (define-public gdb/pinned
   ;; enough to avoid massive rebuilds.
   gdb-11)
 
+(define-public gdb-11.2/pinned
+  ;; This is the fixed version that packages depend on.  Update it rarely
+  ;; enough to avoid massive rebuilds.
+  gdb-11.2)
+
 (define-public gdb
   ;; The "default" version.
   gdb-12)
diff --git a/gnu/packages/rust.scm b/gnu/packages/rust.scm
index d07f4ef10c..3cc5a4926b 100644
--- a/gnu/packages/rust.scm
+++ b/gnu/packages/rust.scm
@@ -788,7 +788,9 @@ (define-public rust
                     (format #f "prefix = ~s" (assoc-ref outputs "rustfmt"))))
                  (invoke "./x.py" "install" "rustfmt")))))))
       ;; Add test inputs.
-      (native-inputs (cons* `("gdb" ,gdb/pinned)
+      (native-inputs (cons* `("gdb" ,(if (target-x86?)
+                                         gdb/pinned
+                                         gdb-11.2/pinned))
                             `("procps" ,procps)
                             (package-native-inputs base-rust))))))
 

[bug#63008] gdb on aarch64

[Andreas Enge]

Hello Pierre,

Am Mon, Apr 24, 2023 at 08:04:18PM +0100 schrieb Pierre Langlois:
> I think the main reason we still have gdb 11 is to make sure we don't
> rebuild the rust world. I'm afraid if we don't do update it know, we'll
> have to do it quite soon, can the rust world be rebuilt on a staging
> branch soon after the core-updates merge?

definitely, there already is a rust-team branch, and the rust team is in
the starting block. Discussing gdb in this context would probably be
a good idea!

Andreas

[bug#63008] gdb on aarch64

[Efraim Flashner]

[-- Attachment #1: Type: text/plain, Size: 1484 bytes --]

On Mon, Apr 24, 2023 at 10:12:35PM +0200, Andreas Enge wrote:
> Hello Pierre,
> 
> Am Mon, Apr 24, 2023 at 08:04:18PM +0100 schrieb Pierre Langlois:
> > I think the main reason we still have gdb 11 is to make sure we don't
> > rebuild the rust world. I'm afraid if we don't do update it know, we'll
> > have to do it quite soon, can the rust world be rebuilt on a staging
> > branch soon after the core-updates merge?
> 
> definitely, there already is a rust-team branch, and the rust team is in
> the starting block. Discussing gdb in this context would probably be
> a good idea!

I also noticed that gdb-11.1 failed to build on core-updates on
riscv64-linux. I can confirm that gdb-12 does build on riscv64-linux on
core-updates. I haven't tried yet to use gdb-12 in place of gdb-11 in
the rust tests, but I assume it would work. I'll test it out on x86_64
on core-updates to see if it works.

It sounds like as it currently stands neither aarch64 nor riscv64 have a
working rust on core-updates. I suppose, in order to decrease the number
of rebuilds, we could introduce gdb-11.2 and use that as the gdb input
in rust for those two architectures until I can either switch everyone
to 11.2 in the rust-team branch or switch everyone to gdb-12.

-- 
Efraim Flashner   <efraim@flashner.co.il>   אפרים פלשנר
GPG key = A28B F40C 3E55 1372 662D  14F7 41AA E7DC CA3D 8351
Confidentiality cannot be guaranteed on emails sent or received unencrypted

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[bug#63008] gdb on aarch64

[Andreas Enge]

Hello Efraim,

Am Tue, Apr 25, 2023 at 01:46:47PM +0300 schrieb Efraim Flashner:
> It sounds like as it currently stands neither aarch64 nor riscv64 have a
> working rust on core-updates.

I am not sure whether it works, but at least it builds on aarch64:
   https://ci.guix.gnu.org/build/37421/details
This is rust-1.60, our publicly visible version; internally we are at 1.65,
which I suppose the rust-team branch is going to make visible soon.
I also see a few green dots of rust-* packages in the dashboard, alongside
many red dots; a random sample of them shows they suffer from the infamous
cuirass bug "cannot build missing derivation".

So I think we are good as far as core-updates is concerned, and you can
make the changes on the rust-team branch when you see fit.

Andreas

[bug#63008] gdb on aarch64

[Andreas Enge]

Am Tue, Apr 25, 2023 at 03:05:51PM +0200 schrieb Andreas Enge:
> I am not sure whether it works, but at least it builds on aarch64:
>    https://ci.guix.gnu.org/build/37421/details
> This is rust-1.60

It was also the version from January, while the current one does not build.
Sorry for the noise, and looking forward to the rust-team branch merge!

Andreas

[bug#63008] gdb on aarch64

[Efraim Flashner]

[-- Attachment #1: Type: text/plain, Size: 1006 bytes --]

On Mon, Apr 24, 2023 at 10:12:35PM +0200, Andreas Enge wrote:
> Hello Pierre,
> 
> Am Mon, Apr 24, 2023 at 08:04:18PM +0100 schrieb Pierre Langlois:
> > I think the main reason we still have gdb 11 is to make sure we don't
> > rebuild the rust world. I'm afraid if we don't do update it know, we'll
> > have to do it quite soon, can the rust world be rebuilt on a staging
> > branch soon after the core-updates merge?
> 
> definitely, there already is a rust-team branch, and the rust team is in
> the starting block. Discussing gdb in this context would probably be
> a good idea!

I was able to build rust-1.60 on core-updates on x86_64 using
gdb-minimal. IMO we should switch to that for aarch64/riscv64 and then
I'll switch x86_64 over to it later in the rust-team branch.

-- 
Efraim Flashner   <efraim@flashner.co.il>   אפרים פלשנר
GPG key = A28B F40C 3E55 1372 662D  14F7 41AA E7DC CA3D 8351
Confidentiality cannot be guaranteed on emails sent or received unencrypted

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

bug#63008: gdb on aarch64

[Efraim Flashner]

[-- Attachment #1: Type: text/plain, Size: 1199 bytes --]

On Tue, Apr 25, 2023 at 03:01:15PM +0300, Efraim Flashner wrote:
> On Mon, Apr 24, 2023 at 10:12:35PM +0200, Andreas Enge wrote:
> > Hello Pierre,
> > 
> > Am Mon, Apr 24, 2023 at 08:04:18PM +0100 schrieb Pierre Langlois:
> > > I think the main reason we still have gdb 11 is to make sure we don't
> > > rebuild the rust world. I'm afraid if we don't do update it know, we'll
> > > have to do it quite soon, can the rust world be rebuilt on a staging
> > > branch soon after the core-updates merge?
> > 
> > definitely, there already is a rust-team branch, and the rust team is in
> > the starting block. Discussing gdb in this context would probably be
> > a good idea!
> 
> I was able to build rust-1.60 on core-updates on x86_64 using
> gdb-minimal. IMO we should switch to that for aarch64/riscv64 and then
> I'll switch x86_64 over to it later in the rust-team branch.

I've since switched rust to using gdb-12.1 and removed gdb-11 from Guix.
Closing this bug!

-- 
Efraim Flashner   <efraim@flashner.co.il>   רנשלפ םירפא
GPG key = A28B F40C 3E55 1372 662D  14F7 41AA E7DC CA3D 8351
Confidentiality cannot be guaranteed on emails sent or received unencrypted

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]