Combine data/text sections: Pad sections to alignment (#197)

* Combine data/text sections: Pad all sections to 4-byte minimum alignment

* Update x86 test snapshot

* Read and store object section alignment

* Combine data/text sections: Pad sections to more than 4-byte alignment if they have alignment specified

Author: LagoLunatic

objdiff-core/src/obj/mod.rs

@@ -9,7 +9,10 @@ use alloc::{
     vec,
     vec::Vec,
 };
-use core::{fmt, num::NonZeroU32};
+use core::{
+    fmt,
+    num::{NonZeroU32, NonZeroU64},
+};
 
 use flagset::{FlagSet, flags};
 
@@ -70,6 +73,7 @@ pub struct Section {
     pub kind: SectionKind,
     pub data: SectionData,
     pub flags: SectionFlagSet,
+    pub align: Option<NonZeroU64>,
     pub relocations: Vec<Relocation>,
     /// Line number info (.line or .debug_line section)
     pub line_info: BTreeMap<u64, u32>,
@@ -105,6 +109,12 @@ impl Section {
         }
     }
 
+    // The alignment to use when "Combine data/text sections" is enabled.
+    pub fn combined_alignment(&self) -> u64 {
+        const MIN_ALIGNMENT: u64 = 4;
+        self.align.map(|align| align.get().max(MIN_ALIGNMENT)).unwrap_or(MIN_ALIGNMENT)
+    }
+
     pub fn relocation_at<'obj>(
         &'obj self,
         obj: &'obj Object,
@@ -363,6 +373,7 @@ static DUMMY_SECTION: Section = Section {
     kind: SectionKind::Unknown,
     data: SectionData(Vec::new()),
     flags: SectionFlagSet::empty(),
+    align: None,
     relocations: Vec::new(),
     line_info: BTreeMap::new(),
     virtual_address: None,

objdiff-core/src/obj/read.rs

@@ -4,7 +4,7 @@ use alloc::{
     string::{String, ToString},
     vec::Vec,
 };
-use core::cmp::Ordering;
+use core::{cmp::Ordering, num::NonZeroU64};
 
 use anyhow::{Context, Result, anyhow, bail, ensure};
 use object::{Object as _, ObjectSection as _, ObjectSymbol as _};
@@ -17,7 +17,7 @@ use crate::{
         Symbol, SymbolFlag, SymbolKind,
         split_meta::{SPLITMETA_SECTION, SplitMeta},
     },
-    util::{read_u16, read_u32},
+    util::{align_data_slice_to, align_u64_to, read_u16, read_u32},
 };
 
 fn map_section_kind(section: &object::Section) -> SectionKind {
@@ -257,6 +257,7 @@ fn map_sections(
             kind,
             data: SectionData(data),
             flags: Default::default(),
+            align: NonZeroU64::new(section.align()),
             relocations: Default::default(),
             virtual_address,
             line_info: Default::default(),
@@ -739,7 +740,10 @@ fn do_combine_sections(
         }
         offsets.push(current_offset);
         current_offset += section.size;
+        let align = section.combined_alignment();
+        current_offset = align_u64_to(current_offset, align);
         data_size += section.data.len();
+        data_size = align_u64_to(data_size as u64, align) as usize;
         num_relocations += section.relocations.len();
     }
     if data_size > 0 {
@@ -754,6 +758,7 @@ fn do_combine_sections(
         let section = &mut sections[i];
         section.size = 0;
         data.append(&mut section.data.0);
+        align_data_slice_to(&mut data, section.combined_alignment());
         section.relocations.iter_mut().for_each(|r| r.address += offset);
         relocations.append(&mut section.relocations);
         line_info.append(&mut section.line_info.iter().map(|(&a, &l)| (a + offset, l)).collect());

objdiff-core/src/obj/snapshots/objdiff_core__obj__read__test__combine_sections.snap

@@ -14,6 +14,7 @@ expression: "(sections, symbols)"
                 8,
             ),
             flags: FlagSet(),
+            align: None,
             relocations: [
                 Relocation {
                     flags: Elf(
@@ -53,6 +54,7 @@ expression: "(sections, symbols)"
                 12,
             ),
             flags: FlagSet(Combined),
+            align: None,
             relocations: [
                 Relocation {
                     flags: Elf(
@@ -87,6 +89,7 @@ expression: "(sections, symbols)"
                 0,
             ),
             flags: FlagSet(Hidden),
+            align: None,
             relocations: [],
             line_info: {},
             virtual_address: None,
@@ -101,6 +104,7 @@ expression: "(sections, symbols)"
                 0,
             ),
             flags: FlagSet(Hidden),
+            align: None,
             relocations: [],
             line_info: {},
             virtual_address: None,

objdiff-core/src/obj/split_meta.rs

@@ -3,6 +3,10 @@ use alloc::{string::String, vec, vec::Vec};
 use anyhow::{Result, anyhow};
 use object::{Endian, ObjectSection, elf::SHT_NOTE};
 
+#[cfg(feature = "std")]
+use crate::util::align_data_to_4;
+use crate::util::align_size_to_4;
+
 pub const SPLITMETA_SECTION: &str = ".note.split";
 pub const SHT_SPLITMETA: u32 = SHT_NOTE;
 pub const ELF_NOTE_SPLIT: &[u8] = b"Split";
@@ -190,17 +194,6 @@ where E: Endian
     }
 }
 
-fn align_size_to_4(size: usize) -> usize { (size + 3) & !3 }
-
-#[cfg(feature = "std")]
-fn align_data_to_4<W: std::io::Write + ?Sized>(writer: &mut W, len: usize) -> std::io::Result<()> {
-    const ALIGN_BYTES: &[u8] = &[0; 4];
-    if len % 4 != 0 {
-        writer.write_all(&ALIGN_BYTES[..4 - len % 4])?;
-    }
-    Ok(())
-}
-
 // ELF note format:
 // Name Size | 4 bytes (integer)
 // Desc Size | 4 bytes (integer)

objdiff-core/src/util.rs

@@ -1,4 +1,4 @@
-use alloc::format;
+use alloc::{format, vec::Vec};
 use core::fmt;
 
 use anyhow::{Result, ensure};
@@ -39,3 +39,24 @@ pub fn read_u16(obj_file: &object::File, reader: &mut &[u8]) -> Result<u16> {
     *reader = &reader[2..];
     Ok(obj_file.endianness().read_u16(value))
 }
+
+pub fn align_size_to_4(size: usize) -> usize { (size + 3) & !3 }
+
+#[cfg(feature = "std")]
+pub fn align_data_to_4<W: std::io::Write + ?Sized>(
+    writer: &mut W,
+    len: usize,
+) -> std::io::Result<()> {
+    const ALIGN_BYTES: &[u8] = &[0; 4];
+    if len % 4 != 0 {
+        writer.write_all(&ALIGN_BYTES[..4 - len % 4])?;
+    }
+    Ok(())
+}
+
+pub fn align_u64_to(len: u64, align: u64) -> u64 { len + ((align - (len % align)) % align) }
+
+pub fn align_data_slice_to(data: &mut Vec<u8>, align: u64) {
+    const ALIGN_BYTE: u8 = 0;
+    data.resize(align_u64_to(data.len() as u64, align) as usize, ALIGN_BYTE);
+}

objdiff-core/tests/snapshots/arch_arm__read_arm.snap

@@ -1519,6 +1519,9 @@ Object {
                 556,
             ),
             flags: FlagSet(),
+            align: Some(
+                1,
+            ),
             relocations: [
                 Relocation {
                     flags: Elf(
@@ -1718,6 +1721,9 @@ Object {
                 0,
             ),
             flags: FlagSet(),
+            align: Some(
+                4,
+            ),
             relocations: [],
             line_info: {},
             virtual_address: None,
@@ -1732,6 +1738,9 @@ Object {
                 76,
             ),
             flags: FlagSet(),
+            align: Some(
+                1,
+            ),
             relocations: [
                 Relocation {
                     flags: Elf(
@@ -1883,6 +1892,9 @@ Object {
                 0,
             ),
             flags: FlagSet(),
+            align: Some(
+                4,
+            ),
             relocations: [],
             line_info: {},
             virtual_address: None,
@@ -1897,6 +1909,9 @@ Object {
                 0,
             ),
             flags: FlagSet(),
+            align: Some(
+                4,
+            ),
             relocations: [],
             line_info: {},
             virtual_address: None,
@@ -1911,6 +1926,9 @@ Object {
                 0,
             ),
             flags: FlagSet(),
+            align: Some(
+                1,
+            ),
             relocations: [],
             line_info: {},
             virtual_address: None,
@@ -1925,6 +1943,9 @@ Object {
                 0,
             ),
             flags: FlagSet(),
+            align: Some(
+                1,
+            ),
             relocations: [],
             line_info: {},
             virtual_address: None,