Sorry, I already fixed my problem. It was related to my Apple Silicon chip
and docker.
If anyone face this issue, I solved it by running docker like this:
FROM *--platform=linux/amd64* ubuntu:latest
Thanks,
Matias
On Wednesday, September 17, 2025 at 5:10:55 PM UTC-4 Mati Cohen wrote:
> Sorry, in addition to my above question:
>
> I've started to set up my environment to build v8 (14.0.263.4).
>
> When I run the following command:
>
> local arch=$1
> local output_dir=$2
>
> local config=()
>
> # building for Android on a specific arch
> config+=('target_os="android"')
> config+=("target_cpu=\"$arch\"")
> config+=("v8_target_cpu=\"$arch\"")
>
> # release configuration
> config+=('is_debug=false')
> config+=('symbol_level=0')
> config+=('android_unstripped_runtime_outputs=false')
>
> # static library
> config+=('is_component_build=false')
> config+=('v8_static_library=true')
> config+=('v8_monolithic=true')
>
> # use builtin snapshotting
> config+=('v8_use_external_startup_data=false')
>
> # disable i18n to reduce size
> config+=('v8_enable_i18n_support=false')
>
> # log everything to stdout on Android
> config+=('v8_android_log_stdout=true')
>
> # starting from v8 9.9 custom libc compiler causes overlaping issues with
> libstdc++.
> # we should explicitly disable it
> config+=('use_custom_libcxx=false')
>
> tools/dev/v8gen.py -vv -b android.arm.release "$output_dir" -- "${config[@
> ]}"
>
> It fails with the following message:
>
> /mnt/upstream/v8/buildtools/linux64/gn gen out.gn/arm64-v8a --check
> -> returned 1
> ERROR at //gni/snapshot_toolchain.gni:115:1: Assertion failed.
> assert(v8_snapshot_toolchain != "",
> ^-----
> Do not know how to build a snapshot for
> //build/toolchain/android:android_clang_arm on linux arm64
> See //BUILD.gn:17:1: whence it was imported.
> import("gni/snapshot_toolchain.gni")
> ^----------------------------------
> GN gen failed: 1
>
> And I also get this (before):
>
> Writing """\
> android_unstripped_runtime_outputs = false
> dcheck_always_on = false
> default_min_sdk_version = 21
> is_debug = false
> target_cpu = "arm"
> target_os = "android"
> v8_android_log_stdout = true
> """ to /mnt/upstream/v8/out.gn/arm64-v8a/args.gn.
>
> It looks like my arguments are being ignored. However, if I go to
> *v8/infra/mb/mb_config.pyl
> *and I hardcode my flags in the android mixin it works properly.
> Is there a bug or am I running something wrong?
>
> Thanks,
> Mati.
>
>
> On Tuesday, September 16, 2025 at 12:53:10 PM UTC-4 Mati Cohen wrote:
>
>> I'm willing to recompile the latest stable version. I am looking at
>> build_config.h and for it's using the following implementation:
>>
>> #if (defined(V8_OS_MACOS) && defined(V8_HOST_ARCH_ARM64)) || \
>> (defined(V8_OS_ANDROID) && \
>> (defined(V8_HOST_ARCH_ARM64) || defined(V8_HOST_ARCH_X64))) || \
>> defined(V8_HOST_ARCH_LOONG64) || defined(V8_HOST_ARCH_MIPS64) || \
>> defined(V8_OS_IOS)
>> // Android 64 bit has experimental support for 16kB pages.
>> // MacOS & iOS on arm64 uses 16kB pages.
>> // LOONG64 and MIPS64 also use 16kB pages.
>> constexpr int kMinimumOSPageSize = 16 * 1024;
>> #elif defined(V8_OS_LINUX) && !defined(V8_OS_ANDROID) && \
>> (defined(V8_HOST_ARCH_ARM64) || defined(V8_HOST_ARCH_PPC64))
>> // Linux on arm64 (excluding android) and PPC64 can be configured for up
>> to 64kB
>> // pages.
>> constexpr int kMinimumOSPageSize = 64 * 1024;
>> #else
>> // Everything else uses 4kB pages.
>> constexpr int kMinimumOSPageSize = 4 * 1024;
>> #endif
>>
>> Sorry for my ignorance, but does it mean that all Android devices with
>> *arm64-v8a* and *x86_64* architectures will use a size of 16KB. Is this
>> correct? I am trying to understand if there are devices with *arm64-v8a*
>> and *x86_64* architectures which still use 4KB for page sizes.
>>
>> Thanks,
>> Matias.
>>
>> On Saturday, September 13, 2025 at 7:17:56 AM UTC-4 Ben Noordhuis wrote:
>>
>>> Assuming recompiling is not a problem for you: just edit
>>> build_config.h and change it to 16k.
>>>
>>> IIRC, stdout and stderr are not normally hooked up to anything on
>>> Android but it should be possible to freopen them to a log file.
>>>
>>> On Fri, Sep 12, 2025 at 4:15 PM 'Matias Cohen' via v8-users
>>> <[email protected]> wrote:
>>> >
>>> > Thanks Ben.
>>> >
>>> > Yeah, that’s strange. I also noticed that SetDataReadOnly didn’t exist
>>> in 10.4.132. Moreover, the V8_FATAL exception that is raised was added even
>>> later. Maybe my records about 10.4.132 are wrong, but I am sure that we
>>> built v8 last time in August 2022, so I can’t explain it.
>>> >
>>> > Android Studio don’t show any error message more than the stacktrace I
>>> already shared.
>>> >
>>> > My guess is that kMinimumOSPageSize was still 4KB for Android back
>>> then when I built v8 (yet I can’t explain why SetDataReadOnly is part of
>>> the stacktrace).
>>> >
>>> > Is there any way for me to overwrite kMinimumOSPageSize from “outside”
>>> v8, just to test if that makes a difference?
>>> >
>>> > Thanks!
>>> >
>>> > > On Sep 12, 2025, at 5:20 AM, Ben Noordhuis <[email protected]>
>>> wrote:
>>> > >
>>> > > On Fri, Sep 12, 2025 at 8:26 AM 'Mati Cohen' via v8-users
>>> > > <[email protected]> wrote:
>>> > >>
>>> > >> Hi,
>>> > >>
>>> > >> My app is using v8 to run a JS environment. Last time I built v8
>>> was about 3 years ago by pointing to 10.4.132.
>>> > >>
>>> > >> As per Google policy, my app needs to support 16 KB page size and
>>> I've been checking all my dependencies to meet that criteria.
>>> > >>
>>> > >> Apparently, based on my checks, my v8 build does support it and I
>>> managed to run the app on my device.
>>> > >> However, v8 is crashing as soon as I call V8::Initialize(); with
>>> the following stack trace:
>>> > >>
>>> > >> /base.apk!liben8.so (offset 0x8ad8000) (v8::base::OS::Abort()+28)
>>> > >> /base.apk!liben8.so (offset 0x8ad8000) (V8_Fatal(char const*,
>>> ...)+376)
>>> > >> /base.apk!liben8.so (offset 0x8ad8000)
>>> (v8::base::OS::SetDataReadOnly(void*, unsigned long)+52)
>>> > >> /base.apk!liben8.so (offset 0x8ad8000)
>>> (v8::internal::V8::Initialize()+1552)
>>> > >> /base.apk!liben8.so (offset 0x8ad8000) (v8::V8::Initialize(int)+28)
>>> > >> /base.apk!liben8.so (offset 0x8ad8000)
>>> (Runtime::Runtime(_JNIEnv*)+172)
>>> > >>
>>> > >> As explained at
>>> https://www.google.com/url?q=https://developer.android.com/guide/practices/page-sizes%23check-code&source=gmail-imap&ust=1758273654000000&usg=AOvVaw2eBFhXRa0E0P-iWsikoJCW,
>>>
>>> I was trying to look for references to PAGE_SIZE and 4096 to see if I was
>>> setting up something wrong, but I didn't find anything.
>>> > >>
>>> > >> Does anyone have any knowledge about this?
>>> > >> Thanks a lot,
>>> > >> Mati
>>> > >
>>> > > V8 doesn't use PAGE_SIZE, it uses sysconf(_SC_PAGESIZE).
>>> > >
>>> > > You mention 10.4.132 but that stack trace is from a different
>>> version,
>>> > > isn't it? Because SetDataReadOnly didn't exist back then.
>>> > >
>>> > > It crashes when trying to mprotect(PROT_READ) the command line
>>> flags,
>>> > > v8_flags. v8_flags is aligned to kMinimumOSPageSize and that
>>> _should_
>>> > > be 16k on Android but you'll want to double-check that. It's defined
>>> > > in src/base/build_config.h.
>>> > >
>>> > > What error message did V8 print?
>>> > >
>>> > > --
>>> > > --
>>> > > v8-users mailing list
>>> > > [email protected]
>>> > >
>>> https://www.google.com/url?q=http://groups.google.com/group/v8-users&source=gmail-imap&ust=1758273654000000&usg=AOvVaw3Vi2w6FFQXBkM9c-JZXWPp
>>>
>>> > > ---
>>> > > You received this message because you are subscribed to a topic in
>>> the Google Groups "v8-users" group.
>>> > > To unsubscribe from this topic, visit
>>> https://www.google.com/url?q=https://groups.google.com/d/topic/v8-users/7XDxoIaXdV4/unsubscribe&source=gmail-imap&ust=1758273654000000&usg=AOvVaw0flrzsS8NN-9t4czOPfrqp.
>>>
>>>
>>> > > To unsubscribe from this group and all its topics, send an email to
>>> [email protected].
>>> > > To view this discussion visit
>>> https://www.google.com/url?q=https://groups.google.com/d/msgid/v8-users/CAHQurc-SN0whK9Y3U8%253DSEdMPOVJLfhQ7fVPcqUi_kAjDJywGeQ%2540mail.gmail.com&source=gmail-imap&ust=1758273654000000&usg=AOvVaw0W3G7qAhJMvkfEdoop6uqR.
>>>
>>>
>>> >
>>> > --
>>> > --
>>> > v8-users mailing list
>>> > [email protected]
>>> > http://groups.google.com/group/v8-users
>>> > ---
>>> > You received this message because you are subscribed to the Google
>>> Groups "v8-users" group.
>>> > To unsubscribe from this group and stop receiving emails from it, send
>>> an email to [email protected].
>>> > To view this discussion visit
>>> https://groups.google.com/d/msgid/v8-users/D27EC591-82E7-43AB-B309-D7D01E9B3123%40encircleapp.com.
>>>
>>>
>>>
>>
--
--
v8-users mailing list
[email protected]
http://groups.google.com/group/v8-users
---
You received this message because you are subscribed to the Google Groups
"v8-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion visit
https://groups.google.com/d/msgid/v8-users/9cb54670-26c6-4c63-ad67-942d21aeb388n%40googlegroups.com.
